RAGE, Receptor of Advanced Glycation Endoproducts, Negatively Regulates Chondrocytes Differentiation

RAGE, receptor for advanced glycation endoproducts (AGE), has been characterized as an activator of osteoclastgenesis. However, whether RAGE directly regulates chondrocyte proliferation and differentiation is unclear. Here, we show that RAGE has an inhibitory role in chondrocyte differentiation. RAGE expression was observed in chondrocytes from the prehypertrophic to hypertrophic regions. In cultured cells, overexpression of RAGE or dominant-negative-RAGE (DN-RAGE) demonstrated that RAGE inhibited cartilaginous matrix production, while DN-RAGE promoted production. Additionally, RAGE regulated Ihh and Col10a1 negatively but upregulated PTHrP receptor. Ihh promoter analysis and real-time PCR analysis suggested that downregulation of Cdxs was the key for RAGE-induced inhibition of chondrocyte differentiation. Overexpression of the NF-κB inhibitor I-κB-SR inhibited RAGE-induced NF-κB activation, but did not influence inhibition of cartilaginous matrix production by RAGE. The inhibitory action of RAGE was restored by the Rho family GTPases inhibitor Toxin B. Furthermore, inhibitory action on Ihh, Col10a1 and Cdxs was reproduced by constitutively active forms, L63RhoA, L61Rac, and L61Cdc42, but not by I-κB-SR. Cdx1 induced Ihh and Col10a1 expressions and directly interacted with Ihh promoter. Retinoic acid (RA) partially rescued the inhibitory action of RAGE. These data combined suggests that RAGE negatively regulates chondrocyte differentiation at the prehypertrophic stage by modulating NF-κB-independent and Rho family GTPases-dependent mechanisms.


Introduction
Advanced glycation end products (AGEs) are permanently modified protein derivatives formed in the presence of reducing sugars, such as glucose and fructose by non-enzymatic glycation, oxidation and dehydration reactions [1]. In diabetic vascular complications including bone disease, AGEs are known to accumulate in various tissues at an extremely accelerated rate [2,3]. 3 classes of AGE receptors: RAGE (receptor for AGE), a complex of OST-48/80KH/galectin-3, class A scavenger receptor have been identified [4][5][6][7][8]. RAGE is a type I transmembrane protein belonging to the immunoglobulin superfamily and is composed of an extracellular region, a hydrophobic transmembrane-spanning domain and a short cytoplasmic tail [5]. Deletion of the cytoplasmic domain of RAGE imparts a dominant negative (DN)-effect on RAGE-dependent activation of cell signaling both in vitro and in vivo [9][10][11]. RAGE signaling axes ultimately induce the nuclear translocation of Nuclear factor (NF)-kB, a hallmark of the pro-inflammatory signal transduction cascade [12]. RAGE is involved in a broad range of inflammatory, degenerative and hyper proliferative diseases, including sepsis, rheumatoid arthritis, diabetic nephropathy/angiopathy, atherosclerosis, cancer and neurological disorders [13,14]. RAGE is expressed in a range of cell types, including smooth muscle cells, fibroblasts, osteoblasts, and osteoclasts [15][16][17][18]. Recent studies using the knockout strategy demonstrated that RAGE, via regulating osteoclast maturation and activation, acts as a bone modulator [15,19]. RAGE null mice showed following phenotypes: increased bone mass and bone mineral density, enhanced bone biomechanical strength and decreased osteoclastic bone resorptive activity. Osteoclasts from RAGE null mice exhibited disrupted actin ring and sealing zone structures, impaired differentiation and attenuated bone resorption activity [15]. Osteoclasts are regulated by bone-forming cells such as osteoblasts and stromal cells. In hard tissue, the accumulation of AGEs by crosslinking in collagen fibrils contributes to disturbed bone modeling and deterioration of bone tissue quality [20]. AGEs-dependent fragility of the bone alters bone mechanical properties such as stiffness and strength [21]. In addition, RAGE is also expressed in articular chondrocytes and it may mediate AGEs-induced osteoarthritis [22]. In the human articular cartilage, an increase in AGE levels negatively affects the proteoglycan synthesis, thereby reducing cellular turnover and repair capacity in turn contributing to the degradation of tissue [23]. These observations on skeletalgenesis led us to the hypothesis that RAGE might directly modulate chondrocyte functions such as proliferation or differentiation.
With respect to RAGE signaling, several target genes have been identified in the past, including proinflammetory mediators, matrix mataroproteinases and adhesion molecules. However, their expression critically depends on cell type, microenvironment and quality of the stimulus [24]. Additionally, although multiple intracellular signaling pathways, including MAP kinases, Rho GTPases, PI3K, JAK/STAT, and NF-kB, have been found to be altered following RAGE stimulation, the molecular mechanisms on how RAGE triggers intracellular signaling to regulate cellular decisions remain largely elusive and the identity of direct signaling molecules downstream of the receptor to modulate chondrocytes are still unknown [25][26][27].
During early skeletal development, mesenchymal cells condense and acquire the chondrocyte phenotype including ability to produce Col2a1 and proteoglycan. In the process of endochondral ossification, immature chondrocytes proliferate and chondrocytes at the center of the cartilaginous skeleton begin to mature to become prehypertrophic chondrocytes which express parathyroid hormone/parathyroid hormone-related peptide receptor (PTHrP-R) and Indian hedgehog (Ihh). The prehypertrophic chondrocytes further mature to hypertrophic chondrocytes that express Col10a1. Upon the terminal differentiation terminal hypertrophic chondrocytes express osteopontin, the matrix is mineralized, vascular vessels invade the calcified cartilage and finally the cartilage is replaced by bone. Chondrocyte proliferation and differentiation occur in an organized manner and result in the formation of a growth plate that is composed of layers of chondrocytes at different stages of differentiation, including resting, proliferating, prehypertrophic, hypertrophic and terminal hypertrophic chondrocytes [28]. However, it is not clear how RAGE influences chondrocyte maturation.
Existing data suggests that accumulated AGE predisposes development of metabolic bone diseases such as osteoarthritis, rheumatoid arthritis. In order to clarify the precise role of AGE-RAGE signaling axis, we studied effects of overexpression of RAGE or DN-RAGE on proliferation, matrix synthesis and differentiation in chondrocytes. Here, we report that RAGE negatively mediated chondrocyte differentiation at prehypertrophic stage thorough NF-kB-independent and Rho family GTPases-dependent mechanisms.

Materials and Methods
Cell cultures and retroviruses ATDC5, MC3T3-E1 cells were purchased from RIKEN Cell Bank (Tsukuba Science City, Japan) and cultured as described previously [29]. To produce the retrovirus, Plat-E cells were transfected with pMXs-neo-derived vectors [30] by Fugene 6 (Roche Diagnostics, Tokyo, Japan). For infection, cells were incubated in the culture supernatant of Plat-E transfectants with 4 mg/ml polybrene for 24 hours and then selected using antibiotic G418 (400 mg/ml). Adenoviruses amplified in 293 cells described below were used as a supernatant aliquot after centrifugation and stored at 285uC until use. NF-kB inhibitor I-kB-SR adenovirus, which has alanine substitutions at serines 32 and 36 was gifted from Dr. Jun-ichiro Ionue [31]. Toxin B and HMGB1 were purchased from Sigma (Sigma Chemical Co., Saint Luis, MO). Primary chondrocytes were prepared as described previously [32]. Briefly, isolated skeleton from E13.5 embryo was dispersed by PBS containing 0.1% trypsin/0.1% collagenase. The protocol used here meets the guideline of the Japanese Society for Pharmacology and was approved by the Committee for Ethical Use of Experimental Animals at Ritsumeikan University, permit number (BKC2010-3-3, BKC2013-018).

Immunocytochemical and immunohistochemical analysis
Detection of alkaline phosphatase activity was performed as described previously [29]. For immunocytochemical analysis, antibodies were reacted before fixation. Cell cultured in chamber slides were reacted with RAGE antibody (AB9714; 1:1000, Millipore, Billerica, MA), and washed by PBS, then fixed by 10% neutralized formaldehyde. Then, samples were reacted with FITC-conjugated anti-rabbit antibody and 1 mg/ml DAPI (49, 6diamidino-2-phenylindole dihydrochloride). For histological analyses, legs of embryos at E15.5 were fixed in 4% paraformaldehyde/phosphate buffered saline, and the limb was embedded in paraffin as described previously and several modifications [29]. Sections (7 mm thick) were subjected to immunohistochemical analysis using antibodies of anti-RAGE (1:500) or anti-Cdx1 (ab116111; 1:500, Abcam, Tokyo, Japan). Antibodies reactions were performed in Blocking One Histo (Nacalai Tesque, Kyoto, Japan). Fluorescence images were acquired using EVOS FL cell imaging system (Life Technologies Corp., Tokyo, Japan). Transmit and green mode images were overlaid and obtained merged images.

AGE preparation
AGE preparation conducted as described by Tanaka et al. [33]. 50 mg/ml of bovine serum albumin (Sigma Chemical Co.) was incubated under sterile conditions with 0.5 M glucose, 1.5 mM PMSF, 0.5 mM EDTA and antibiotics for 6 weeks in phosphatebuffered saline (pH7.4). Non-glycated BSA (BSA) was incubated under same conditions except for the absence of glucose as a negative control. The incorporated glucose was removed by dialysis against PBS using dialysis tube C-75 (Viskase Companies Inc., Darien, IL). After dialysis, BSA and AGE-BSA (AGE) were concentrated by using a filtration device (VIVASPIN 500, 10,000 MWCO PES, Sartorius, Germany). Protein concentrations were determined with a BCA protein assay kit (Sigma Chemical Co.).

Cell proliferation assay
Cell proliferation activity was measured using a colorimetric Cell Count Reagent SF kit (Nacalai Tesque) according to manufacturer's instruction. Cells were plated in 96-well plates at a density of 3,000 cells/well (ATDC5) or 10,000 cells/well (MC3T3-E1). Cells were treated with BSA or AGE-BSA for 2 days. After cells were incubated with WST-8 for 2 hours, proliferative activities were measured on a microplate reader at 450 nm (model680, Bio-Rad, Tokyo, Japan). There was no difference in the number of dead cells between the cell lines determined by a trypan blue exclusion assay.

Cell cycle analysis
Each gene transferred cells (1,000,000 cells) in 145 mm dish were analyzed using Millipore Cell Cycle Detection kit (EMD Millipore Corp. Hayward, CA). Cells were incubated for 180 min with fixer, then washed by phosphate buffered saline. Cell cycles of each cells were detected in MUSE cell analyzer (EMD Millipore Corp.).

Quantitative Real-time PCR
Total RNA was extracted using Sepazol (Nacalai Tesque). One mg of total RNA was reverse-transcribed by ReverTra Ace cDNA synthesis kit (TOYOBO). For quantitative real-time PCR, 5 ml of 2 x KAPA master mix (Nippon Genetics Co. ltd., Tokyo, Japan) was mixed with cDNA, 0.5 mM primer pair. Triplicate of cDNA's were amplified on Piko Real PCR system (Thermo Fisher Scientific Inc., Yokohama, Japan). The experiments were performed at five different cDNA pool dilutions. PCR products were normalized against GAPDH, and measurements between samples were compared by cycling threshold (Ct). Primer sequences used are the following: GAPDH-

Micromass culture
For micromass culture, limb buds from E12.5 embryos were isolated and were digested in 0.1% trypsin/0.1% collagenase for 30 min at 37uC as described previously [32]. Briefly, cells were suspended by pipetting and reaction was stopped by Dulbecco's Modified Eagle Medium (DMEM) (WAKO Pure Chemical Industries, Ltd., Osaka, Japan) containing 10% fetal bovine serum (FBS). Cells were resuspended in 20 ml drops at 2610 7 cells/ml, plated to 12-well plates. After 1 hour to allow the attachment of cells, 2 ml of DMEM containing 10% FCS were overlaid. After 24-h, chemical treatment or virus infection were started. And then, alucian blue staining was performed to detect the production of cartilaginous matrix as described previously [32]. Medium was changed every day. For alucian blue staining, the wells were rinsed twice with PBS and fixed with methanol for 5-min. They were then stained overnight with 0.1% alucian blue in 20% acetic acid/ 80% ethanol. Wells were rinsed three times with distilled water and dye was extracted with 6 M guanidine-HCl. Absorbance of extracted dye was measured at 620 nm.

Chromatin immunoprecipitation
Chromatin immunoprecipitation was performed as described previously [34]. Briefly, GFP control cells or Flag-tagged Cdx1 cells were fixed by paraformaldehyde and fragmented genome were prepared by sonication (Bronson sonifier model 150, setting 5, 5sec-25-times). After immunoprecipitation using anti-Flag M2 beads (Sigma Chemical Co.), DNA was purified from supernatants and immunoprecipitates. Real-time PCR was performed using the following primers specific to the promoter region of mouse Ihh gene: Ihh-Cdx-F, 59-TGGATTGGGATCCGTTTGTT-39; Ihh-Cdx-R, 59-AAGCAAACCCGAAAGCTGGA-39. From the experiment using anti-HA antibody (Santa Cruz Biotechnology, Inc., Y-11), we confirmed the specificity of pull-down by M2 beads.

Statistical analysis
Data are expressed as mean 6S.E.M. Significance was tested using a Student's t-test or, where multiple comparisons were required, Tukey-test or Williams-test. A P-value of less than 0.05 was considered to be significant.

The expression of RAGE in embryo skeleton
AGEs are known as RAGE ligands and exert mitogenic action in vascular smooth muscle cells [33]. We first examined RAGE immunoreactivity in primary chondrocytes and chondrogenic ATDC5 cells. Following antibody reaction before fixation, RAGE was observed around the cell membrane in chondrocytes (Fig.1A, B). RAGE expression levels in primary chondrocytes and ATDC5 cells were similar (Fig.1C). To elucidate physiological expression of RAGE in in vivo, we tested RAGE immunoreactivity in embryonic hind limbs (Fig.1D, E). RAGE was mainly expressed from the prehypertrophic to hypertrophic region and perichondorium in E15.5 hind limbs. Reactivity against RAGE antibody was only weakly observed in proliferative chondrocytes, and negligibly in cancellous bone, periosteum and bone collar. These results suggested that RAGE regulates chondrocyte in the prehypertrophic to hypertrophic stages.

RAGE-independent ATDC5 proliferation
Using prepared BSA or AGE-BSA (AGE), we first examined the effect of AGE on ATDC5 proliferation. AGE significantly promoted ATDC5 proliferation in a concentration-dependent manner (Fig.1F). On the other hand, AGE had no effect on osteoblastic MC3T3-E1 cells (Fig.S1G). In attempt to examine the dependency of RAGE on chondrocyte proliferation, stable ATDC5 transfected cells expressing GFP, RAGE or DN-RAGE were established using retroviruses (Fig.1G-J). Mitogenic action of AGE was not influenced by overexpression of RAGE or DN-RAGE compared to GFP control cells (Fig.1G). RAGE expression levels in RAGE or DN-RAGE cells were clearly elevated (Fig.1H). In addition, basal cell cycle status of these cells was unaffected (Figl.S1A-F). Although slightly, the progression into S phase was inhibited by RAGE. These results suggested that AGEs induced ATDC5 proliferation in a RAGE-independent manner. A previous report demonstrated that AGEs activate human RAGE promoter [33]. Therefore using stable transfected cells of RAGE or DN-RAGE, we examined the effect of AGE on human RAGE promoter activities (Fig.1I). Elevated RAGE promoter activity by AGE was tended to facilitate by RAGE. Basal and AGE-induced RAGE promoter activities were inhibited by DN-RAGE significantly. Using a specific primer pair to detect endogenous RAGE, real-time PCR analysis demonstrated that RAGE by RAGE overexpression with or without AGE was significantly upregulated, and down regulated by DN-RAGE (Fig.1J). These results suggested that extracellular AGE stimulation influences chondrocyte function, DN-RAGE functions as dominant negative form and that AGE regulated RAGE expression.

The role of RAGE on chondrocytes differentiation
To monitor differentiation state of stable transfected cells of GFP, RAGE or DN-RAGE, we performed ALP staining ( Fig.2A). ALP transcripts in each stable transfected cells were not influenced (data not shown). Next, we performed micromass culture to monitor cartilaginous matrix production. After isolation of limb buds from E12.5 mouse embryos, cells were infected with GFP-Ad, RAGE-Ad, or DN-RAGE-Ad. Two days later after infection, cells were stained by alucian blue (Fig.2C). The results showed that RAGE-Ad inhibited cartilaginous matrix production and DN-RAGE-Ad augmented production compared to GFP-Ad control. To investigate RAGE action on gene regulation, mRNA expression levels were analyzed by real-time PCR. In ATDC5 cells, Sox9, Col1a1, Col2a1, Runx2, PTHrP, VEGF, MMP-13, osteopontin and osteocalcin expressions were not changed by RAGE-Ad or DN-RAGE-Ad, while Ihh, col10a1 were downregulatd by RAGE-Ad and upregulated by DN-RAGE-Ad ( Fig.2E-I, K-P). Conversely, PTHrP receptor (PTHrP R) was significantly upregulated by RAGE-Ad (Fig.2J). This data strongly suggested that RAGE negatively regulated chondrocyte differentiation at the prehypertrophic stage.

NF-kB-independent and Rho family GTPases-dependent mechanisms
To examine dependency on AGE of RAGE effects on cartilaginous matrix production, we tested the effect of AGE in GFP-, RAGEand DN-RAGE-infected cells using adenoviruses (Fig.3A). Results showed that AGE did not influence cartilaginous matrix production in comparison to BSA treatment in all infected cells. On the other hand, one candidate ligand for RAGE, HMGB1 clearly inhibited them in a concentration-and RAGEdependent manner (Fig.S3A, B). To clarify the involvement of NF-kB signal that is known to be promoted by RAGE, we assessed NF-kB activity. For this, we established NF-kB-luc stable transfected cells of ATDC5. NF-kB activity was stimulated by RAGE-Ad but not by DN-RAGE-Ad (Fig.3B). HMGB1 stimulated NF-kB activity in GFP-Ad transfected cells, but not in DN-RAGE-Ad transfected cells (Fig.S3C). I-kB inhibitor, I-kB-SR-Ad effectively inhibited basal and RAGE-induced NF-kB activities. I-kB-SR-Ad did not affect cartilaginous matrix production in both GFP-Ad and RAGE-Ad (Fig.3C). Additionally, the inhibitory effects on Ihh and col10a1 mRNA expressions by RAGE-Ad were not influenced by I-kB-SR-Ad (Fig.3G, H). These results suggested that RAGE inhibited chondrocyte differentiation in a NF-kB-independent manner.
Next, we investigated downstream RAGE signaling. Recent report on chondrocyte maturation signals suggests that there is significant cross-talk among the pathways and that the overall effects on chondrocyte function is dependent on the balance in activity of multiple signaling proteins [37]. We first tested Toxin B, a Rho family GTPases inhibitor action. As shown in Fig.3F, Toxin B solely did not influence cartilaginous matrix production and restored RAGE-Ad-induced inhibitory action. In reporter analysis, Toxin B restored RAGE-Ad-induced inhibitory effects on Ihh and Col10a1 promoter activities without influencing NF-kB activity  Fig.3 D, E, and data not shown). We next analyzed effect of the activation state of Rho family GTPases. We established stable transfected cells of constitutively active forms of Rho family GTPases, L63RhoA, L61Rac or L61Cdc42 in ATDC5 by infection using retroviruses. Real-time PCR analysis showed decreased Ihh and Col10a1 mRNA expressions in L63RhoA, L61Rac, or L61Cdc42 (Fig.3I, J). These results suggested that RAGE inhibited chondrocyte differentiation in a Rho family GTPases-dependent manner.

Inhibitory mechanism of chondrocyte differentiation by RAGE
To explore the precise mechanism of chondrocyte differentiation regulated by RAGE, we analyzed inhibition of Ihh expression by RAGE using Ihh promoter deletional constructs [34]. RAGEinduced inhibitory action and DN-RAGE-induced stimulatory action were maintained in p1312-luc and p994-luc, while these actions were disrupted in p597-luc (Fig.4A). To determine RAGEdependent responsive DNA elements within region from 2994 to 2597 in Ihh promoter, we performed in silico anaysis using TF Search program (http://www.cbrc.jp/research/db/TFSEARCH. html) (Fig.S2). From transcription factor binding sites within 397 bp of Ihh promoter sequence, Cdxs were identified as candidate factors modulated by RAGE. Real-time PCR analysis demonstrated that Cdxs expression was inhibited by RAGE-Ad and stimulated by DN-RAGE-Ad (Fig.4B-D). To examine the dependency on the NF-kB pathway, we tested effect of I-kB-SR-Ad on RAGE-induced Cdxs downreduction (Fig.4E-G). I-kB-SR-Ad did not influence Cdxs expression with or without RAGE-Ad. In addition, L63RhoA, L61Rac or L61Cdc42 showed decreased Cdxs expressions in comparison to GFP control (Fig.4H-J).

Characterization of Cdxs as a target molecule of RAGE signaling
We examined Cdx1 expression pattern in in vivo (Fig.5A, B). Cdx1 immunoreactivity was abundant in prehypertrophic chondrocytes and poor in proliferative and hypertrophic chondrocytes and the perichondorium. On the other hand, no expression was observed in cancellous bone, periosteum and bone collar at E15.5. Next, to determine the direct interaction of Cdx1 and Ihh promoter, we performed chromatin immunoprecipitation (ChIP) assay. For this, we established stable transfected cells of MOCK or Flag-tagged Cdx1 in ATDC5 cells. Approximately, a ten-fold increase of Cdx1 expression was observed in Flag-tagged Cdx1 cells (Fig.5F). ChIP analysis showed direct interaction between Cdx1 and Ihh promoter (Fig.5C). Transient overexpression studies demonstrated that Cdx1 stimulated Ihh p994-luc and col10a1 p3000-luc promoter activities (Fig.5D, E). In stable transfected cells, Cdx1 upregulated Ihh and Col10a1 expressions compared to MOCK control (Fig.5G, H). Additionally, Cdx1-Ad restored RAGE-Ad-induced inhibition of cartilaginous matrix production (Fig.5I). Previous reports showed that Cdx1 is a direct target gene of retinoic acid (RA) [38]. Therefore we examined whether or not RA would affect RAGE inhibitory action. RA rescued RAGEinduced inhibition of cartilaginous matrix production in a concentration dependent fashion (Fig.6A, B). 0.1 mM RA exerted protective action on cartilaginous matrix production inhibited by RAGE-Ad. RA also induced Cdx1 expression transiently (Fig.6C). We examined the difference of RA response using GFP and RAGE stable transfected cells (Fig.6D). Immunoblot analysis showed that Cdx1 expression was reduced in RAGE cells compared to GFP cells without RA treatment, and Cdx1 induction by 0.1 mM RA was clearly observed in RAGE cells. RA itself did not influence Ihh promoter activity (Fig.6E). Reduced Ihh promoter activity caused by RAGE-Ad was partially restored after 0.1 mM RA treatment.

Discussion
Damage to the cartilage is a major problem, especially in joint disease. AGEs are one of the candidates that act as adverse mediators in joint diseases [22,23]. RAGE has been shown to be an initiator of inflammatory response in immune responsive cells such as macrophages [39]. Although RAGE knockout mice show skeletal abnormalities after birth [15,19], the biological importance of RAGE expression in chondrocytes remains to be clarified. Hence in this work, we examined the precise mechanisms, 1; regulatory mechanism of chondrocytes functions by RAGE, 2; involvement of most conceivable inflammatory NF-kB or Rho family GTPases signals in chondrocytes. We showed for the first time that RAGE activation in chondrocytes functions as a mediator for suppression of chondrocyte differentiation via NF-kB-independent and Rho family GTPases-dependent mechanisms.

RAGE as a negative regulator of cell differentiation
Previous reports showed that AGEs stimulate human vascular endothelial cell proliferation [33]. As there was a possibility that AGEs action might be cell type specific, we examined AGEs action on chondrocyte proliferation (Fig.1F). Overexpression studies demonstrated that RAGE did not function as mitogen in chondrocytes (Fig.1G), suggesting that the mitogenic action of AGEs is mediated via receptor(s) other than RAGE. Immunohistochemical analysis showed that RAGE expression was mainly in prehypertrophic stage chondrocytes and the perichondorium at E15.5 (Fig.1D, E). Furthermore, AGEs did not influence to MC3T3-E1 proliferation (Fig.S1G). These results suggested that during fetal development RAGE was functional in chondrocytes but not in osteoblasts. On the other hand, knockout mice of Hmgb1, an endogenous candidate ligand for RAGE, had abnormal skeletal phenotype during fetal development [26,27]. These observations suggest that Hmgb1 probably controls skeletal formation in a RAGE-independent manner during fetal development. As our results indicated that AGE slightly influences proliferation but not cartilaginous matrix production (Fig.1F,  Fig.3A), and AGE influenced RAGE expression (Fig.1I, J), exogenous HMGB1 clearly inhibited cartilaginous matrix production (Fig.S3), suggesting that AGE somewhat influences chondrocytes function, HMGB1 may function as key factor in some pathological or physiological conditions. Taking all this into consideration, we hypothesize that AGE accumulation is as a result of physical flexibility and stiffness as opposed to receptor activation by bioactive molecules. Further investigation for the role of AGE-RAGE signaling axis will be required.
We examined RAGE function in the absence of ligand stimulation since RAGE overexpression solely evoked significant receptor activation as shown in Fig.3B. This could be attributed to stimulation by FCS-derived factors. The effect of AGE as a RAGE ligand was limited in our systems as described above. In attempt to unravel how RAGE regulates chondrocytes, we designed overexpression experiments studies using RAGE and DN-RAGE. Pathologically, RAGE is expressed in articular chondrocytes and is thought to mediate AGE-induced osteoarthritis [22]. In human articular cartilage, an increase in AGE levels negatively affects proteoglycan synthesis thereby reducing cellular turnover and repair capacity hence contributing to tissue degradation [23]. We assumed from the pathologic high expression of RAGE after birth that RAGE does play a functional role in chondrocytes. RAGE clearly inhibited cartilaginous matrix production (Fig.2C), influenced expression of two prehypertrophic chondrocytes differentiation markers, Ihh and PTHrP receptors and the hypertrophic chondrocyte differentiation marker Col10a1 (Fig.2J, K, N). These results demonstrated that even without AGE stimulation, We also demonstrated that Cdxs identified from the Ihh promoter by in silico analysis, upregulated Ihh and Col10a1 expressions ( Fig.4A-D, Fig.5G, H). The Cdx (caudal) gene family of homeodomain transcription factors, Cdx1, Cdx2, and Cdx4, are thought to act as modulators of vertebral axial patterning [40]. Cdx1 and Cdx2 are expressed in the primitive streak region at E7.5, followed by Cdx4 expression at E8.5 [41]. Spatio-temporally regulated Cdx expression has been proposed to be indicative of a functional Cdx gradient that regulates spatial expression of target genes along the major body axis [42,43]. Cdxs are regulated by Homeobox genes, which modulate spatio-temporal gene expression to form precise body axis [44][45][46][47]. It is considerable that Cdxs are downregulated by RAGE, subsequently suppressing the expression of Ihh and Col10a1. Indeed, Cdxs gradient is important in the formation of the body axis [41], and RAGE disturbed Cdxs expression levels in cultured cells at least. Metabolic cartilaginous diseases cause deformation in joints. Excessive RAGE accumulation in joint diseases has been reported [23,24]. This suggests that pathological RAGE activation probably causes disturbance of Cdxs, in turn leading to skeletal deformation. Based on these observations, we investigated the effect of Cdx1 overexpression on RAGE-induced cartilaginous matrix production and on gene regulation (Fig.4B-D). Cdx1 was expressed from the prehypertrophic to hypertrophic region of chondrocytes an expression pattern similar to that of RAGE (Fig.1D, E, Fig.5A, B). Additionally, RAGE overexpression downregulated Cdx1 (Fig.4B-D). DN-RAGE overexpression clearly inverted gene expressions, suggesting that Ihh, Col10a1, and Cdxs were RAGE-sensitive genes. These processes were all NF-kB-independent (Fig.3C, G, H and Fig.4E-G). The downstream RAGE signals include small GTPases Rho family protein functions as mediators [25]. RAGE inhibitory action was restored by the Rho family GTPases inhibitor Toxin B (Fig.3D-F) while constitutively active forms of all Rho GTPases downregulated Cdxs, Ihh and Col10a1 (Fig.3I, J, Fig.4H-J). In addition, Cdx1 clearly facilitated chondrocyte differentiation (Fig.5G-I). Altogether these findings suggests the RAGE-Rho GTPases signaling axis functions as a down regulator for Ihh by modulating Cdxs in prehypertrophic stage chondrocytes.

Regulation of spatiotemporal Ihh signal by RAGE
Previous studies on osteoclasts regulation by RAGE showed that RAGE functions as a transducer for osteoclast maturation signal [15,19]. Long bones in skeletal parts by endochondral ossification were elongated in RAGE knockout mice compared to wild type mice after birth. Elongated long bone length observed in RAGE knockout mice is suggestive of the possibility of additional RAGE downstream signals besides the osteoclast regulation pathway. Ihh is an essential factor for cartilaginous longitudinal development and RAGE activation resulted in Cdxs-Ihh downregulation (Fig.2K, Fig.4B-D). On the other hand, the disruption of RANKL, an essential osteoclast stimulator, caused dwarfism by osteopetrosis [48]. The binding of RANKL to its receptor RANK, promotes osteoclast maturation and activation in the NF-kB pathway. While, RAGE-Ad stimulated NF-kB in chondrocytes (Fig.3B), I-kB-SR-Ad did not influence gene regulation of Ihh and Col10a1 (Fig.3G, H). Additionally, we studied whether RAGE affects the RANKL and OPG expressions or not (Fig.S4A, B). Interestingly, RAGE failed to regulate RANKL and OPG expressions. AGE also failed to stimulate RANKL and OPG promoter activities (Fig.S4C, D). Therefore, our results cue a new aspect of RAGE action in chondrocyte differentiation, and osteoclasts activation by RAGE would be independent on RANKL and OPG.
As described above, Cdxs came up as the candidates for RAGE stress indicator. Cdx1 was induced by RA transiently (Fig.6C). Although RA did not influence to cartilaginous matrix production or Ihh promoter activities without RAGE stress, RA responsiveness was observed only in stress condition (Fig.6A, B, E). Basal Cdx1 protein expression levels were clearly reduced in RAGEtransduced cells compared to GFP cells, and Cdx1 was induced in response to RA even in RAGE-transduced cells (Fig.6D). Therefore, we suggested that Cdx1 induction by RA may transiently work only in stress conditions such as excessive RAGE activation. We preliminarily examined Cdxs protein stability and intracellular localization using GFP-Cdx1 transduced cells, and speculated from the data that RA treatment reduced Cdxs degradation hence increasing utilization efficiency of Cdxs. Further investigation will be required to clarify these observations and are currently in progress.

Summary
In this study, we investigated the mechanisms of cartilaginous deterioration by RAGE, and found a pivotal role of Cdxs as the target of excessive RAGE activation state. Classically, antiinflammation drugs such as NSAIDs or steroids are insufficient to control cartilaginous metabolic diseases such as osteoarthritis and rheumatoid arthritis. Deformation in metabolic cartilaginous disease remains a serious problem. We suggest the possibility of therapeutic effectiveness of RA in such conditions. We intend to proceed with further verification of RA effectiveness in the treatment of metabolic cartilaginous conditions with deformations. Figure S1 Cell cycle regulation by RAGE in ATDC5 and insufficiency of cell proliferation by AGE in MC3T3-E1. (A-F) Each stable transfected cells at approximately 70% confluences in 145 mm dishes were fixed, and stained by propidium iodide according to manufactures instruction. Cell cycle analysis showed that RAGE reduces slightly G2/M phase, and increased G0/G1 phase compared to GFP. Furthermore, RAGE did not affect the cells in S phase. On the other hand, DN-RAGE had no effect on cell cycle. G, AGE did not stimulate cell proliferation in osteoblastic MC3T3-E1. n = 8-16.