Activation of the receptor EphB4 by its specific ligand ephrin B2 in human osteoarthritic subchondral bone osteoblasts




Abnormal subchondral bone metabolism is involved in osteoarthritis (OA). It has been suggested that ephrin B2 and its specific receptor EphB4 participate in bone homeostasis. We previously reported that human OA subchondral bone osteoblasts could be classified into 2 subpopulations: low (L), having proresorption properties, and high (H), having proformation properties. The purpose of this study was to investigate the importance of the ephrin system in OA subchondral bone osteoblasts.


The presence of the EphB4 receptor was determined by immunohistochemistry, and its expression level, modulation upon treatment, and consequences of activation by ephrin B2 were determined by quantitative polymerase chain reaction. The effects of ephrin B2 activation of the EphB4 receptor on bone resorption activity were also determined. EphB4 receptor activation signaling pathways were investigated by specific enzyme-linked immunosorbent assay.


EphB4 receptors were present in subchondral bone osteoblasts and osteocytes. Compared with normal and H-OA osteoblasts, EphB4 receptor expression levels were significantly increased in L-OA osteoblasts, with no difference between normal and H-OA osteoblasts. EphB4 receptor levels in L-OA osteoblasts were significantly up-regulated by prostaglandin E2 (PGE2) and interleukin-17 (IL-17). Ephrin B2, PGE2, and IL-17 significantly inhibited bone resorption activity in these cells. EphB4 activation by ephrin B2 significantly inhibited the expression of IL-1β, IL-6, matrix metalloproteinase 1 (MMP-1), MMP-9, MMP-13, and RANKL, but not MMP-2 and osteoprotegerin. EphB4 receptor activation significantly inhibited the phosphatidylinositol 3-kinase/Akt pathway.


This study is the first to provide evidence that EphB4 receptor activation by ephrin B2 in OA subchondral bone could affect abnormal metabolism in this tissue by inhibiting resorption factors and their activities. Ephrin B2 could be targeted as a specific therapeutic approach in the development of a disease-modifying OA drug.

Subchondral bone is made up of a specialized connective tissue formed by a mineralized matrix containing the specific type I collagen, proteoglycans, and several growth factors and cytokines, as well as bone-specific cell types (osteoblasts, osteocytes, and osteoclasts) (1, 2). Osteoblast and osteoclast activities, either alone or in combination, contribute to the bone remodeling process, and any disturbance in the activities of these 2 cells is responsible for the development of altered bone metabolism.

The ephrin B molecules (ephrin B1, ephrin B2, and ephrin B3) bind in a specific manner to their EphB receptors (EphB1 through EphB6) and to some EphA receptors (3–6). Both the ephrins and the Eph receptors are membrane-bound proteins, and their interaction leads to bidirectional (osteoblast/osteoclast) Eph/ephrin signaling. In this context, signaling through the EphB receptors is considered forward signaling and through the ephrin B ligands, reverse signaling (3–6). Although the ephrin systems are known to play a crucial role in the development of several tissues and organs, including the nervous and cardiovascular systems (7–9), a new and potent role in bone biology has recently been proposed. Osteoclasts express only ephrin B1 and B2, without any detectable EphB receptors, whereas osteoblasts express both of these ephrin B ligands and the EphB receptors (5). Interestingly, it was recently demonstrated that ephrin B2, which is expressed by osteoclasts, and its specific receptor EphB4, which is expressed by osteoblasts, are involved in the control of bone homeostasis. More specifically, their interaction leads through EphB4 forward signaling to an osteoblast differentiation process, whereas reverse signaling through ephrin B2 ligands leads to an inhibitory effect on the osteoclast function. The overall outcome of such interaction favors bone formation (5).

OA is characterized mainly by degradation of cartilage, inflammation of synovial membranes, and important changes in subchondral bone. More than 30 years ago, Radin et al (10, 11) suggested that changes in bone might be a cause of OA. Since that time, there has been substantial evidence that changes in the metabolism of bone, particularly in the area of the subchondral bone, are an integral part of the disease (12–18). Recent studies, however, point to the fact that the fate of articular cartilage is not determined exclusively by stiffening (sclerosis) of subchondral bone, but rather, by a remodeling of this tissue (19, 20). Some clinical studies performed in OA patients have shown that the markers of bone resorption are increased early in the disease course (21, 22), whereas subchondral bone sclerosis is a relatively late phenomenon. Taken together, the data suggest that abnormal subchondral bone metabolism occurs at an early stage of the disease and is the driving force behind the degradation and loss of cartilage.

Alterations in the OA subchondral bone are numerous and involve morphologic and biochemical changes in osteoblasts as well as structural alterations of the tissue. Several studies have provided strong evidence that during OA, phenotype differences in subchondral bone osteoblasts are present, and these cells are metabolically more active. Chondrocytes, osteoblasts, and synovial fibroblasts are the first source of enzymes responsible for progression of the disease, and it is widely accepted that members of the matrix metalloproteinase (MMP) family play a major role (23). Moreover, considerable evidence indicates that proinflammatory cytokines synthesized and released by bone cells contribute either directly or indirectly to the process of osteoclastogenesis and, hence, are crucial to the metabolic alterations in OA, affecting the development/progression of the disease.

It is also well known that bone metabolism is tightly controlled by a molecular triad composed of osteoprotegerin (OPG)/receptor activator of nuclear factor κB (NF-κB)/RANK ligand (RANKL). This triad has been well established for its capacity to control osteoclast biology (24). RANKL is produced by osteoblasts and is essential for inducing bone resorption. RANKL stimulates osteoclastogenesis and osteoclast activity by binding to the cell surface receptor RANK, which is located on osteoclast precursors and mature osteoclasts (24). Such binding leads to the activation of specific signaling pathways involved in the formation and survival of osteoclasts and in bone resorption (25, 26). OPG secreted by osteoblasts acts as a soluble decoy receptor for RANKL. Hence, and as demonstrated in our previous study (20), RANKL and OPG appear to be of great importance in the control of the subchondral bone remodeling process in humans.

In previous studies, we showed that human OA subchondral bone osteoblasts could be discriminated into 2 subpopulations: low (L) and high (H) osteoblasts (27, 28). L-OA osteoblasts showed bone proresorption activities, and H-OA osteoblasts showed proformation properties (20). There is, to our knowledge, no study to date on the possible implication of ephrin during the course of OA. We therefore investigated the presence and modulation of EphB4 receptors in human subchondral bone osteoblasts, as well as the functional consequences of the activation of this receptor by its endogenous ligand ephrin B2. We found that normal, L-OA, and H-OA subchondral bone osteoblasts differentially express the EphB4 receptor and that treatment with prostaglandin E2 (PGE2) and interleukin-17 (IL-17) favors an increased level of EphB4 receptor expression. Interestingly, activation of EphB4 by ephrin B2 induced a reduced bone remodeling process through inhibition of various mediators of catabolism, including the inflammatory factors IL-1β and IL-6, MMP-1, 13, and 9, and RANKL. This appears to occur through an inhibition of the phosphatidylinositol 3-kinase (PI 3-kinase)/Akt signaling pathway.


Specimen selection.

Human subchondral bone was obtained from the femoral condyles of OA patients (n = 32) undergoing total knee arthroplasty (mean ± SD age 71 ± 9 years) or normal subjects (n = 3) within 12 hours of death (mean ± SD age 65 ± 16 years). OA was evaluated according to American College of Rheumatology clinical criteria (29). At the time of surgery, the OA patients had symptomatic disease requiring medical treatment in the form of acetaminophen, nonsteroidal antiinflammatory drugs, or selective cyclooxygenase 2 inhibitors. None had received intraarticular steroid injections within the 3 months prior to surgery. None of the normal subjects or OA patients had received medication that would interfere with bone metabolism. The institutional Ethics Committee Board of the University of Montreal Hospital Centre approved the use of human articular tissues for this study.

Subchondral bone osteoblast culture.

Cultures of subchondral bone osteoblasts were prepared as previously described (27). Briefly, bone samples were cut into small pieces and digested for 4 hours with type I collagenase in BGJb medium (both from Sigma-Aldrich Canada, Oakville, Ontario, Canada), without serum, at a temperature of 37°C in a humidified atmosphere of 5% CO2/95% air. The bone pieces were then cultured in BGJb medium containing 20% heat-inactivated fetal bovine serum (FBS; PAA Laboratories, Etobicoke, Ontario, Canada) and an antibiotic mixture (100 units/ml of penicillin base and 100 μg/ml of streptomycin base) (Multicell; Wisent, St. Bruno, Quebec, Canada) at 37°C in a humidified atmosphere of 5% CO2/95% air. When cells were observed in the Petri dishes, the culture medium was replaced with fresh medium containing 10% FBS, and incubation continued until the cells were confluent. Osteoblasts were used after 1 passage.

For determination of EphB4 receptor and ephrin B2 ligand expression levels in normal and OA cells, RNA was extracted (see below) as soon as the cells reached confluence. The effects of the osteotropic factors on EphB4 receptor levels and ephrin B2 activation of the EphB4 receptor were assessed by preincubating cells in Dulbecco's modified Eagle's medium (DMEM; Gibco BRL, Burlington, Ontario, Canada)/0.5% FBS for 24 hours, followed by 18 hours of incubation with fresh DMEM/0.5% FBS containing the factors under study. The factors tested were IL-1β (100 pg/ml; Genzyme, Cambridge, MA); tumor necrosis factor α (TNFα; 5 ng/ml), PGE2 (500 nM), IL-6 (10 ng/ml), and IL-17 (10 ng/ml) (all from R&D Systems, Minneapolis, MN); and ephrin B2 ligand (50 and 100 ng/ml; Abnova, Walnut, CA). The concentrations were chosen based on previous studies or studies reported in the literature.

RNA extraction, reverse transcription, and real-time polymerase chain reaction (PCR).

Total cellular RNA from human osteoblasts was extracted with TRIzol reagent (Invitrogen, Burlington, Ontario, Canada) according to the manufacturer's specifications. RNA was quantified using a RiboGreen RNA quantification kit. The reverse transcription reactions were primed with random hexamers as previously described (30).

Real-time quantification of messenger RNA was performed as previously described (30), using a Rotor-Gene 6 RG-3000A real-time PCR system (Corbett Research, Mortlake, New South Wales, Australia) with 2× QuantiTect SYBR Green PCR Master Mix (Qiagen, Mississauga, Ontario, Canada) according to the manufacturer's specifications.

The primer sequences used were as follows: for EphB4 receptor, 5′-CACAGTCATCCAGCTCGTG-3′ (antisense) and 5′-ATCGGATGGGAATCTTTCC-3′ (sense); for ephrin B2 ligand, 5′-TTCGACAACAAGTCCCTTTG-3′ (antisense) and 5′-CGAGTGCTTCCTGTGTCTC-3′ (sense); for IL-1β, 5′-CCTGTACGATCACTGAACTG-3′ (antisense) and 5′-TGGGCAGACTCAAATTCCAG-3′ (sense); for IL-6, 5′-CACCTCTTCAGAACGAATTG-3′ (antisense) and 5′-CTAGGTATACCTCAAACTCC-3′ (sense); for MMP-1, 5′-CTGAAAGTGACTGGGAAACC-3′ (antisense) and 5′-AGAGTTGTCCCGATGATCTC-3′ (sense); for MMP-2, 5′-CACTGTTGGTGGGAACTCAG-3′ (antisense) and 5′-GTGTAAATGGGTGCCATCAG-3′ (sense); for MMP-9, 5′-CCTTCACTTTCCTGGGTAAG-3′ (antisense) and 5′-CCATTCACGTCGTCCTTATG-3′ (sense); for MMP-13, 5′-CTTAGAGGTGACTGGCAAAC-3′ (antisense) and 5′-GCCCATCAAATGGGTAGAAG-3′ (sense); for OPG, 5′-GTTTACTTTGGTGCCAGG-3′ (antisense) and 5′-GCTTGAAACATAGGAGCTG-3′ (sense); for RANKL, 5′-GGGTATGAGAACTTGGGATT-3′ (antisense) and 5′-CACTATTAATGCCACCGAC-3′ (sense); and for GAPDH, 5′-CAGAACATCATCCCTGCCTCT-3′ (antisense) and 5′-GCTTGACAAAGTGGTCGTTGAG-3′ (sense).

Data were expressed as the threshold cycle (Ct) and were calculated as the ratio of the number of molecules of the target gene to the number of molecules of GAPDH. Primer efficiencies for the test genes were the same as for the GAPDH gene. Standard curves were generated with the same plasmids used for the target sequences.

Identification of osteoblast subpopulations.

Identification of human OA subchondral bone osteoblast subpopulations was performed as previously described (20, 27, 28). OA osteoblasts producing low levels of PGE2 (≤2,000 pg/mg of protein) were classified as L-OA, and those producing high levels of PGE2 (>2,000 pg/mg of protein) as H-OA.

PGE2 levels were determined in the culture media after 48 hours of incubation, using an enzyme immunoassay (Cayman Chemicals, Ann Arbor, MI) with a sensitivity of 7.8 pg/ml. All determinations were performed in duplicate for each cell culture.

Immunohistochemical analysis.

Subchondral bone specimens were processed for immunohistochemical analysis. Briefly, undecalcified subchondral bone explants were embedded in methylmethacrylate as previously described (31). Sections (5 μm) prepared from the embedded specimens were placed on Superfrost Plus slides (Fisher Scientific, Nepean, Ontario, Canada), deplasticized, rehydrated, and then immunostained. The specimens were washed in phosphate buffered saline (PBS), incubated in 0.3% Triton X-100 for 20 minutes, and placed in 2% hydrogen peroxide/PBS for 15 minutes. Slides were further incubated with a blocking serum (Vectastain ABC assay; Vector, Burlingame, CA) for 45 minutes, after which they were blotted and then overlaid with the primary antibody (15 μg/ml of goat anti-human EphB4 receptor; R&D Systems) for 18 hours at 4°C. The slides were washed 3 times in PBS, pH 7.4, and incubated with the secondary antibody (anti-goat IgG; Vector) for 1 hour at room temperature, followed by staining according to the avidin–biotin–peroxidase complex method (Vectastain ABC assay).

Color was developed with 3,3′-diaminobenzidine (Dako Diagnostics, Mississauga, Ontario, Canada) containing hydrogen peroxide. Slides were counterstained with eosin. All incubations were performed in a humidified chamber. Sections were examined with a Leitz Orthoplan light microscope (Leica, St. Laurent, Quebec, Canada). Two control procedures were performed according to the same experimental protocol: omission of the primary antibody and substitution of the primary antibody with an autologous preimmune serum. Controls showed only background staining.

Signaling pathway determination.

Levels of the phosphorylated MAP kinases (ERK-1/2, p38, SAPK/JNK), NF-κB, and PI 3-kinase/Akt were determined by a cellular activation of signaling ELISA (CASE) assay (SuperArray Bioscience, Frederick, MD) according to the manufacturer's recommendations. This method uses a cell-based enzyme-linked immunosorbent assay (ELISA) that directly measures protein phosphorylation on cultured cells. The CASE assay includes a complete antibody-based detection system for colorimetric quantification of the relative amount of phosphorylated protein. Briefly, following treatment with ephrin B2 ligand at 100 ng/ml for different time periods (5, 15, 30, and 60 minutes), the cells were fixed to preserve any activation-specific protein modification. The procedure was performed according to the manufacturer's specifications (SuperArray Bioscience), and the amount of bound antibody was determined using a developing solution and an ELISA plate reader. The absorbance readings were normalized to the relative cell number, as determined with a cell staining solution according to the manufacturer's recommendation.

Determination of resorption activity.

Resorption activity was measured using the BD BioCoat Osteologic Bone Cell Culture system (BD Biosciences, Oakville, Ontario, Canada) as described previously (20). Briefly, human peripheral blood mononuclear cells (PBMCs; 100,000 cells/well) were inoculated into the wells with culture medium containing DMEM/10% FBS, antibiotics, and 25 ng/ml of macrophage colony-stimulating factor (M-CSF; R&D Systems) and were incubated for 3 days at 37°C in order to induce preosteoclastic differentiation (32). Human OA subchondral bone osteoblasts (10,000 cells/well) were then inoculated with the differentiated PBMCs (preosteoclasts) and incubated for another 3 days. At the end of this period, cells were incubated for 4 weeks at 37°C with fresh DMEM containing M-CSF, 10% FBS, and antibiotics with the factors under study. Media were changed every 3 days. At the end of the incubation period, cells were bleached (6% NaOCl, 5.2% NaCl) and extensively washed in sterilized water. As a contrast stain for resorption, von Kossa's stain was used as described by BD Biosciences. Quantification was performed by light microscopy using Bioquant Osteo II software (version 8.00.20; Bioquant Image Analysis, Nashville, TN). Results were calculated as the percentage of the resorbed surface and were expressed as the percentage of control, where the control was assigned a value of 100%.

Statistical analysis.

Data are expressed as the mean ± SEM. Statistical significance was assessed by Student's 2-tailed t-test. P values less than or equal to 0.05 were considered significant.


Ephrin B2 ligand and EphB4 receptor expression.

The data showed first that human OA subchondral bone osteoblasts had a higher level of ephrin B2 ligand expression than did normal subchondral bone osteoblasts (Figure 1A), but the difference was not statistically significant. We previously (20, 27, 28) demonstrated that the endogenous PGE2 levels in OA subchondral bone osteoblasts discriminated 2 subgroups of OA patients. One subgroup, in which endogenous osteoblast PGE2 levels were comparable to normal, was classified as low producers, and the other subgroup, in which high levels of PGE2 were observed, was classified as high producers. The L-OA osteoblasts were shown to have proresorption activities and the H-OA osteoblasts to have proformation properties (20).

Figure 1.

Levels of ephrin B2 ligand and EphB4 receptor expression in normal and osteoarthritic (OA) human osteoblasts. Ephrin B2 ligand (A and B) and EphB4 receptor (C and D) expression was determined in normal (n = 3) and OA (n = 8) subchondral bone osteoblasts (A and C), as well as in OA osteoblasts producing either low (L) levels of prostaglandin E2 (PGE2) (n = 3) or high (H) levels of PGE2 (n = 5) (B and D). The levels of both ephrin B2 and EphB4 receptor are expressed as a factor of 10−4 arbitrary units. Values are the mean and SEM. P values were determined by Student's 2-tailed t-test.

We next determined the PGE2 levels in OA osteoblasts and then further discriminated each subgroup according to the data. As expected, the data revealed 2 subgroups of osteoblasts: L-OA osteoblasts with mean ± SEM PGE2 levels of 868 ± 148 pg/mg of protein and H-OA osteoblasts with PGE2 levels of 4,498 ± 303 pg/mg of protein. When we further examined the L-OA and H-OA osteoblast subgroups, a statistically significant increase was found for the ephrin B2 ligand on L-OA osteoblasts (n = 3) as compared with normal (n = 3) (P < 0.02) and H-OA (n = 5) (P < 0.004) osteoblasts (Figure 1B). Levels of ephrin B2 ligand expression were similar in normal and H-OA osteoblasts.

EphB4 receptor expression levels were also investigated, and the data showed a trend similar to that for the ephrin B2 ligand: increased expression of the EphB4 receptor in OA compared with normal osteoblasts (Figure 1C). In our examination of the L-OA and H-OA osteoblast subgroups, we found a significant increase in the EphB4 receptor on L-OA osteoblasts (n = 3) as compared with normal (n = 3) (P < 0.002) and H-OA (n = 5) osteoblasts (P < 0.0007) (Figure 1D).

EphB4 receptor protein.

To verify the presence of the EphB4 receptor protein in OA subchondral bone, an immunohistologic experiment was performed, using a specific antibody directed against the receptor (n = 3). As illustrated in Figure 2, this receptor was produced by OA subchondral bone cells, and positive staining was seen for both osteoblasts and osteocytes.

Figure 2.

Representative immunohistologic section of osteoarthritic subchondral bone obtained from the femoral condyle, showing localization of the EphB4 receptor. Staining for the EphB4 receptor on osteoblasts (solid arrows) as well as osteocytes (broken arrows) can clearly be seen in the higher-magnification view of the bracketed area of the main tissue section. (Original magnification × 100 [left]; × 250 [right].)

Modulation of EphB4 receptor expression.

Since no differential expression of the EphB4 receptor was observed in the H-OA osteoblasts as compared with normal osteoblasts, we followed up by investigating factors that could be responsible for the up-regulation of EphB4 receptor expression on L-OA cells. To this end, L-OA osteoblasts were treated with osteotropic factors known to modulate osteoblast metabolism, including IL-1β, TNFα, PGE2, IL-6, and IL-17. As shown in Figure 3, levels of EphB4 receptor expression (n = 7) were significantly increased by treatment with PGE2 or IL-17 (P < 0.04 for each comparison), but no true effect was found for treatment with IL-1β, TNFα, or IL-6.

Figure 3.

EphB4 receptor gene expression in osteoarthritic (OA) subchondral bone osteoblasts producing low levels of prostaglandin E2 (PGE2) (proresorption properties). OA osteoblasts (n = 7) were treated for 18 hours in the absence (control) or presence of the osteotropic factors interleukin-1β (IL-1β), tumor necrosis factor α (TNFα), PGE2, IL-6, and IL-17. At the end of the incubation period, total RNA was extracted and processed for real-time polymerase chain reaction analysis. Values are the mean and SEM arbitrary units over that in the control, which was assigned a value of 1. P values were determined by Student's 2-tailed t-test (versus control).

Functional consequences of activation of the EphB4 receptor by ephrin B2.

We further investigated the modulation of various remodeling factors known to be involved in osteoblast physiologic/pathophysiologic processes in L-OA osteoblasts (n = 5) in the presence and absence of ephrin B2 (50 and 100 ng/ml). These factors included IL-1β, IL-6, MMP-1, MMP-2, MMP-9, MMP-13, OPG, and RANKL.

Interestingly, the data revealed that activation of the EphB4 receptor led to a pattern of reduced expression of many of the remodeling factors upon treatment with ephrin B2. Indeed, the proinflammatory cytokines IL-1β and IL-6 were significantly inhibited. The reduction in IL-1β levels was similar with either 50 ng/ml (P < 0.05) or 100 ng/ml (P < 0.02) of ephrin B2 (Figure 4A). For IL-6, a dose-dependent effect was found, with a significant difference reached at 100 ng/ml (P < 0.04) (Figure 4A). MMP-1, MMP-13, and MMP-9, but not MMP-2, were significantly decreased with ephrin B2 treatment (Figure 4B). A dose-dependent effect was seen for MMP-1 and MMP-13. Finally, the bone remodeling factor RANKL, but not OPG, was also significantly inhibited (P < 0.03) upon treatment with 100 ng/ml of ephrin B2, thereby increasing the ratio of OPG to RANKL (Figure 4C).

Figure 4.

Ephrin B2 activation of the EphB4 receptor in osteoarthritic (OA) subchondral bone osteoblasts producing low levels of prostaglandin E2 (proresorption properties), as indicated by the effects on gene expression levels of A, interleukin-1β (IL-1β) and IL-6, B, matrix metalloproteinase 1 (MMP-1), MMP-13, MMP-2, and MMP-9, and C, osteoprotegerin (OPG), RANKL, and the ratio of OPG to RANKL. OA osteoblasts (n = 5) were treated for 18 hours in the absence (control) or presence of ephrin B2 at 50 ng/ml and 100 ng/ml. At the end of the incubation period, total RNA was extracted and processed for real-time polymerase chain reaction analysis. Values are the mean and SEM arbitrary units over that in the control, which was assigned a value of 1 (broken line). P values (versus control) were determined by Student's 2-tailed t-test.

Modulation of L-OA osteoblast resorption activity.

Since our data showed that activation of the EphB4 receptor with ephrin B2 reduced the levels of many remodeling factors, we further investigated whether activation of the EphB4 receptor affects bone resorption activity. Complementary experiments were also performed with the 2 factors that were shown to up-regulate the EphB4 receptor level, PGE2 and IL-17. We found that ephrin B2 (n = 3), PGE2 (n = 5), and IL-17 (n = 4) all markedly reduced bone resorption activity as compared with autologous controls (Figure 5).

Figure 5.

Resorption activity in osteoarthritic subchondral bone osteoblasts producing low levels of prostaglandin E2 (PGE2) (proresorption properties) that were coincubated with human peripheral blood mononuclear cells in the presence of macrophage colony-stimulating factor and the osteotropic factors PGE2, (500 nM; n = 5) and interleukin-17 (IL-17) (10 ng/ml; n = 4), as well as ephrin B2 (100 ng/ml; n = 3). Values are the mean and SEM percentage of the total surface resorbed upon treatment with the factors as compared with control, which was assigned a value of 100%. P values (versus control) were determined by Student's 2-tailed t-test.

Signaling pathways involved in ephrin B2 induction of the EphB4 receptor.

The effects of EphB4 receptor activation by ephrin B2 on the levels of 3 phosphorylated MAP kinases, namely, ERK-1/2, p38, and SAPK/JNK, in L-OA osteoblasts (n = 4), as well as on the levels of NF-κB and PI 3-kinase/Akt, were determined. Activation of the EphB4 receptor in L-OA osteoblasts at 100 ng/ml of ephrin B2 yielded within 15 minutes a decrease in the phosphorylation of PI 3-kinase/Akt, with statistical significance reached at 30 minutes (P < 0.05). The other signaling pathways that included the MAP kinases and NF-κB were not significantly modulated following EphB4 receptor activation (data not shown).


Previous studies have indicated an abnormal resorption process in the subchondral bone of patients with OA (19, 20), thus indicating alterations in osteoblast metabolism in this tissue as a possible target in the development of specific therapeutic strategies. In this context, we explored the effects of EphB4 receptor activation by its endogenous ligand ephrin B2 on human OA subchondral bone osteoblasts. Our data revealed new and important information about the mechanisms by which the ephrins may exert an inhibitory effect on the remodeling process in OA subchondral bone.

Ephrin B molecules bind in a specific manner to their Eph receptors (3–6), resulting in bidirectional Eph/ephrin signaling. Although it is known that forward signaling through EphB4 receptors in osteoblasts induces osteogenic differentiation and that reverse signaling through ephrin B2 ligand suppresses osteoclast differentiation, the effect of ephrin B2 on the EphB4 receptor in the modulation of bone remodeling factors has not previously been explored, and there are no data concerning these effects on human OA subchondral bone osteoblasts.

The first data to emerge from our study revealed that the EphB4 receptor was differentially expressed by normal osteoblasts and each of the OA subpopulations, with significantly increased expression by L-OA subchondral bone osteoblasts as compared with normal and H-OA osteoblasts. Further immunohistologic analyses demonstrated that the EphB4 receptor is produced in OA subchondral bone by osteoblasts and osteocytes, but not by osteoclasts. This finding is consistent with that reported by Zhao et al (5), who showed that the EphB4 receptor was undetectable on osteoclasts. Other emerging data from our study showed that on OA subchondral bone osteoblasts, ephrin B2–induced EphB4 receptor activation down-regulated various bone remodeling factors and that factors that up-regulate the EphB4 receptor (PGE2 and IL-17) or its activation (ephrin B2) markedly diminished bone resorption activity. These data are significant, since information concerning the interaction between these 2 ephrin family members during the OA process could help in the development of a disease-modifying OA drug that would inhibit the resorption/remodeling process in human OA subchondral bone.

Osteoclastogenesis has been described as occurring through tight control by some members of the TNF superfamily (24), involving RANKL and/or RANKL-independent processes. RANKL, a factor synthesized by cells of osteoblast lineage, is essential for mediating bone resorption, through the enhancement of osteoclast differentiation and proliferation after binding to RANK, a receptor located on osteoclast precursors and mature osteoclast cell surfaces. Another important member of the TNF receptor family is OPG, a factor that acts as a decoy receptor and blocks the binding of RANKL to its receptor RANK, thus inhibiting the process of osteoclastogenesis. In this study, we showed that EphB4 receptor activation by ephrin B2 disturbed the equilibrium between OPG and RANKL, significantly decreasing the level of RANKL gene expression but not affecting that of OPG, with the net outcome indicating an increased ratio of OPG to RANKL. Since osteoclastogenesis is closely related to the ratio of these factors, a reduction in the osteoclastogenesis process would be expected to occur.

This hypothesis is consistent with our data showing that ephrin B2 markedly reduced the bone resorption activity of L-OA osteoblasts. Moreover, ephrin B2–induced EphB4 receptor caused a decrease in the expression levels of the cytokines IL-1β and IL-6, which also supports a reduction in osteoclast formation, survival, and activity. Indeed, these cytokines, either alone or in synergy with RANKL, have been demonstrated to potentiate osteoclastogenesis (33–37). More specifically, IL-1β in the presence of RANKL has been shown to prolong the survival of osteoclasts, induce multinucleation of osteoclasts, and stimulate the formation of the actin ring (a functional marker of osteoclasts) (33, 36, 38). In addition, both IL-1β and IL-6 were shown to significantly increase the membranous localization of RANKL on human OA subchondral bone osteoblasts (37), thus sustaining a bone resorption process.

MMP modulation is also closely linked to bone remodeling (39, 40). In this regard, a study using a canine model of experimental OA showed that during the disease, some members of the MMP family are selectively located at the active subchondral bone resorption zone (19). We therefore further investigated the effect of ephrin B2 activation of the EphB4 receptor on the collagenases MMP-1 and MMP-13 and the gelatinases MMP-2 and MMP-9 in human OA subchondral bone osteoblasts. Interestingly, osteoblasts treated with ephrin B2 significantly inhibited the expression of MMP-1, MMP-9, and MMP-13, but not MMP-2. This could have occurred through a direct effect of EphB4 receptor activation or through an indirect effect via inhibition of the cytokine IL-1β, a major factor involved in the up-regulation of MMP in articular joint tissues (41). The inhibition of these MMPs is important, since these proteases are well-known for their induction of bone resorption via the degradation of extracellular matrix components, including the collagen fiber network, as well as other extracellular components, such as fibronectin and aggrecans.

MMP-2 was also observed to be up-regulated during OA. To our knowledge, however, there are very few published data demonstrating a degradative property of MMP-2 in OA subchondral bone osteoblasts. Rheumatoid arthritis, like OA, is characterized by the joint degeneration, but it occurs in a more aggressive manner. Experiments performed in MMP-2–knockout mice and in MMP-9–knockout mice revealed a reduction in the progression of arthritis only in the MMP-9–knockout mice (42). Hence, the much more significant involvement of MMP-9 than MMP-2 in the process of rheumatoid arthritis could also be of major importance in OA. Moreover, it has also been shown that MMP-9 is highly involved in the osteoclastic bone resorption process by facilitating the migration of osteoclasts through proteoglycan-rich matrices (43).

Together, these findings suggest that in human OA subchondral bone, activation of the EphB4 receptor could act at 2 different levels: by limiting the extent of matrix degradation through the inhibition of MMP and by disturbing the osteoclastogenesis process mediated by the bone remodeling factor RANKL as well as proinflammatory cytokines.

Since our data indicated a higher level of EphB4 receptor in osteoblasts during the proresorption phase (L-OA osteoblasts), we further investigated which factors might be responsible for the receptor up-regulation on these cells. Interestingly, upon treatment with various osteotropic factors known to be involved in the pathologic changes of subchondral bone, the EphB4 receptor level was up-regulated only upon treatment with PGE2 or IL-17. Hence, PGE2 and IL-17 could act as a retroactive mechanism by which the EphB4 receptor expression level increases, which in turn, contributes to a switch from a bone resorption phase to a bone formation phase. This hypothesis is substantiated by the findings that both PGE2 and IL-17, as well as activation of the EphB4 receptor by ephrin B2, exhibited a marked reduction in bone resorption activity.

The effects of PGE2 on bone resorption activity could occur through either a RANKL-dependent or a RANKL-independent mechanism during the differentiation of osteoclast precursors. Indeed, in a previous study (20), RANKL was found to be markedly decreased in H-OA osteoblasts (which produce high levels of PGE2) as compared with L-OA osteoblasts (which produce low PGE2 levels) and normal osteoblasts. In addition, in H-OA osteoblasts, inhibition of PGE2 with indomethacin markedly increased the level of RANKL expression. It was also shown that L-OA, but not H-OA, osteoblasts enhanced the formation of osteoclasts (20). A direct effect of PGE2 on osteoclastogenesis is also possible, since Take et al (44) recently demonstrated such an effect on osteoclasts, which occurred through interactions with PGE2 on its specific receptors.

Additional experiments revealed that the mechanism by which EphB4 receptor activation exerts its effect was through the down-regulation of PI 3-kinase/Akt phosphorylation, but not by affecting the activation of the MAP kinases ERK-1/2, SAPK/JNK, and p38, or NF-κB. These findings are consistent with those from studies of other cell types demonstrating the critical role of PI 3-kinase/Akt in the regulation of proinflammatory cytokines and the production of some MMPs (45–48), as well as with other studies in which activation of the PI 3-kinase/Akt pathway was essential in up-regulating proinflammatory cytokine–induced and oncostatin M–induced MMP-9 and MMP-13 (46, 48). In addition, activation of the PI 3-kinase/Akt pathway in osteoblasts has been demonstrated to play a crucial role in promoting the differentiation of these cells by preventing an apoptotic mechanism (47). The PI 3-kinase/Akt pathway was also shown to be essential in controlling IL-6 synthesis in osteoblasts. Takai et al (45) showed that IL-6 synthesis induced by the proinflammatory cytokine TNFα is suppressed by inhibitors of Akt as well as by the PI 3-kinase inhibitors wortmannin and LY294002. Thus, in addition to diminishing osteoblast differentiation, a significant decrease in PI 3-kinase/Akt phosphorylation could lead to a situation that favors a reduction in the levels of bone catabolic effectors.

Our data concerning the possible involvement of these ephrin molecules in the remodeling process of the subchondral bone during OA can be briefly summarized as follows. In L-OA osteoblasts, the EphB4 receptor could be up-regulated by catabolic factors, including PGE2 and IL-17, and then activated by ephrin B2, which would result in a decreased activation of PI 3-kinase/Akt, which in turn, would inhibit the proinflammatory cytokines IL-1β and IL-6, the proteases MMP-1, MMP-9, and MMP-13, as well as RANKL, all of which are involved in the remodeling process in subchondral bone.

The literature suggests that the ephrin B2 ligand is present in membranous form on both osteoblasts and osteoclasts (5), but the interaction occurs through osteoclast–osteoblast contact. However, Hattori et al (49) recently hypothesized that ephrin could be shed and that the protease ADAM10 may be involved in this cleavage. If such an ephrin B2 cleavage process occurs, then in addition to osteoclasts, osteoblasts could be implicated in providing soluble ephrin B2. Such a process would explain our finding of higher ephrin B2 expression levels in human L-OA subchondral bone osteoblasts. If, on the other hand, ephrin B2 is not shed, it is tempting to propose that these cells would produce more of the EphB4 receptor and its specific ligand ephrin B2 in an attempt to regulate the resorption activity of these cells. This could explain the high level of these factors in L-OA osteoblasts as compared with normal osteoblasts. In H-OA osteoblasts, such an increase would not be required, since these cells are in a proformation phase. This would be consistent with the data from the present study showing that in H-OA osteoblasts, the levels of both EphB4 receptor and ephrin B2 are similar to those in normal osteoblasts.

Our findings that human OA subchondral bone osteoblast activation of the EphB4 receptor inhibits some abnormal biochemical pathways as well as resorption activity might also be applied to other osteolytic diseases. Indeed, diseases such as postmenopausal osteoporosis, rheumatoid arthritis, multiple myeloma, and breast cancer demonstrate a bone remodeling process and increased production of RANKL (50, 51), suggesting that activity in favor of bone resorption could then have led to reduced levels of EphB4 receptor and/or ephrin B2. Hence, in these diseases as well as in OA, treatment with ephrin B2 might be of interest in limiting the production of the cytokines and proteases involved in the remodeling process and in preventing the up-regulation of RANKL, and therefore the abnormal osteoclastogenesis.

In conclusion, this study is the first to provide evidence that EphB4 receptor activation by ephrin B2 in human OA subchondral bone could affect the abnormal metabolism in human L-OA osteoblasts by reducing their altered resorption activity. Although much remains to be discovered about this ephrin system, our study brings to light new and important information about the mechanism by which ephrin B2 could exert a protective effect on structural changes in OA articular tissue.


Dr. Martel-Pelletier 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. Kwan Tat, Martel-Pelletier.

Acquisition of data. Kwan Tat, Amiable, Boileau, Duval.

Analysis and interpretation of data. Kwan Tat, Lajeunesse, Pelletier, Martel-Pelletier.

Manuscript preparation. Kwan Tat, Pelletier, Lajeunesse, Martel-Pelletier.

Statistical analysis. Kwan Tat.


The authors are grateful to Martin Boily for preparation of the immunohistologic sections, to Changshan Geng, François-Cyril Jolicoeur, and François Mineau for expert technical assistance with the PCR and cell cultures, and to Virginia Wallis for manuscript preparation.