Introduction

Bone is formed by two processes, intramembranous ossification and endochondral ossification (Fig. 1). In the process of intramembranous ossificiation, osteoblasts differentiated from mesenchymal stem cells directly form bone. On the other hand, in the process of endochondral ossification, chondrocytes differentiated from mesenchymal stem cells first form transient cartilage, the so-called growth-plate cartilage, and then this cartilage is replaced by bone after invasion of blood vessels into the cartilage. Even after that, articular cartilage remains as a permanent cartilage on the surface of joints.

Fig. 1
figure 1

Mechanism of skeletal formation. Bone is formed by two processes, intramembranous ossification and endochondral ossification. Cartilage is classified into two categories, permanent cartilage such as articular cartilage and temporal cartilage, which is involved in endochondral ossification

Full size image

As shown in Fig. 2, in the fetus, bone anlagen is formed as cartilage; and then intramembranous ossification occurs at the center of this anlagen, referred to as the primary ossification center. Then the bone expands in both directions, and a secondary ossification center is formed in the epiphysis. As a result, plate-like cartilage remains between both bones; and this plate-like cartilage is called the growth plate.

Fig. 2
figure 2

(Right) Process of endochondral ossification. Long bones are first formed as cartilage template and then ossification occurs at the center of the cartilaginous bone anlagen, which is called the “primary ossification center.” After a secondary ossification center is formed, the cartilage between primary and secondary ossifications, which is called the growth plate, grows rapidly and is then replaced by bone after invasion of blood vessels into the cartilage. (Left) CCN2 gene was cloned as a hypertrophic chondrocyte specific gene 24 (hcs24). Its gene product promotes proliferation and differentiation of chondrocytes, osteoblasts, and vascular endothelial cells

Full size image

In the growth plate, chondrocytes proliferate, differentiate, mature, and become hypertrophic chondrocytes which form a calcified matrix. Then, osteoclasts erode this calcified cartilage, after which blood vessels penetrate into the eroded cartilage. Also, osteoblasts deposit a bone matrix; and finally the cartilage is replaced by bone. Many growth factors, cytokines, and vitamins have been shown to be involved in chondrocyte proliferation and differentiation, but none of them promotes the entire process of endochondral ossification. To isolate such a factor, we previously carried out differential-display PCR and cloned a gene that was predominantly expressed in hypertrophic chondrocytes. We named it hypertrophic chondrocyte-specific gene hcs 24, the product of which was found to be identical to CTGF, now called CCN2 (Fig. 2; Takigawa 2003; Takigawa et al. 2003; Perbal and Takigawa 2005; Nakanishi et al. 1997).

CCN2 is a member of the CCN protein family, which originally consisted of 3 members, i.e., CTGF, Cyr61 and Nov, and now consists of 6 members (Takigawa et al. 2003; Perbal and Takigawa 2005; Brigstock et al. 2003). These members are cysteine-rich secretory proteins, and each member contains four conserved modules, i.e., an IGFBP module, VWC type C module, TSP type 1 repeat, and C-terminal module except for CCN5, which lacks the CT module (Fig. 3, top). Early studies paid much attention to the involvement of CCN2 in fibrosis, because expression of the CCN2 gene had been shown to be induced by TGF-beta, which expression had been found in various fibrotic disorders, and because CCN2 had been shown to promote the proliferation, migration, and adhesion of fibroblasts (Perbal and Takigawa 2005; Takigawa 2003; Mori et al. 1999; Sato et al. 2000; Brigstock 1999; Moussad and Brigstock 2000). However, the discovery of CCN2 in chondrocytes opened a new era for CCN2 research, especially that regarding its physiological function.

Fig. 3
figure 3

(Top) Modular structure of CCN2. CCN2 consists of the following four conserved modules: insulin-like growth factor binding protein-like (IGFBP) module, Willebrand factor type C (VWC) module, thrombospondin (TSP) type 1 repeat, and C-terminal module. (middle and bottom) Binding partners of CCN2 protein modules and function of each module. Other CCN family proteins in parentheses show that the molecules indicated are their binding partners. FGFR1 binds to CCN2 through other modules except CT. The molecules marked in red were found to bind to CCN2 in the author’s laboratory

Full size image

Role of CCN2 in endochondral ossification - in vitro studies

By making recombinant proteins, my colleagues and I found that CCN2 stimulates the proliferation and differentiation of growth-plate chondrocytes and causes them to become hypertrophic and form a calcified matrix (Nakanishi et al. 2000). We also found that recombinant CCN2 (rCCN2) promotes the proliferation and differentiation of osteoblasts, as well as calcification by them (Nishida et al. 2000). Moreover, we found CCN2 to promote the proliferation, adhesion, migration, and tube formation of vascular endothelial cells and to induce angiogenesis in ovo and in vivo (Shimo et al. 1998, 1999; Kubota and Takigawa 2007a).

From these findings, we hypothesized that CCN2 promotes endochondral ossification by acting on three types of cells: chondrocytes, osteoblasts, and endothelial cells (Takigawa et al. 2003; Kubota and Takigawa 2007b, 2011). Thus we coined the term “Ecogenin: endochondral ossification genetic factor” to describe this molecule (Takigawa et al. 2003). However, there is one more type of cell that plays an important role in endochondral ossification, i.e., the osteoclast. As shown in Fig. 2, when cartilage is replaced by bone, osteoclasts invade into the matrix of the calcified cartilage and provide space for the deposition of osteoid by osteoblasts. Therefore, we recently investigated the effect of CCN2 on osteoclastogenesis.

Using osteoclast precursor cell line RAW 264.7 cells (Shui et al. 2002), we found that (1) CCN2 potentiates RANKL-induced osteoclastogenesis in its late stage; (2) expression of CCN2 and DC-STAMP is induced in the late stage of RANKL-induced osteoclastogenesis, with the induction of CCN2 being earlier than that of DC-STAMP; and (3) CCN2 potentiates the expression of DC-STAMP in the late stage of osteoclastogenesis (Nishida et al. 2011a; Fig. 4). Moreover, using IP-Western blotting and the solid-phase binding assay, we also found that CCN2 binds to DC-STAMP (Nishida et al. 2011a; Fig. 4).

Fig. 4
figure 4

Mechanism of CCN2-enhanced osteoclastogenesis. (Right) CCN2 expressed during osteoclastogenesis promotes osteoclast formation via induction of and interaction with DC-STAMP in the late stage of osteoclastogenesis. In the early stage of osteoclastogenesis, CCN2 interacts with RANK and enhances RANKL signaling, probably due to increased availability of RANKL, because CCN2 also interacts with OPG, a decoy receptor of RANK. The figure was drawn with help by Drs. T. Nishada and E. Aoyama

Full size image

Using fetal liver cells as a source of osteoclast precursor from CCN2 null mice (Nishida et al. 2011a), we also found that osteoclastogenesis induced by M-CSF and RANKL in vitro is impaired in these mice and that exogenous CCN2 rescues this impaired osteoclastogenesis. Furthermore, we showed that knock down of CCN2 by siRNA inhibits pit formation, which is a marker of active osteoclasts, and that this inhibition is reversed by the addition of rCCN2 (Nishida et al. 2011a). Moreover, CCN2 increases DC-STAMP expression in fetal liver cells from CCN2 null mice; and forced expression of DC-STAMP by retrovirus transfer increases the formation of TRAP-positive cells from fetal liver cells (Nishida et al. 2011a). These findings indicate that CCN2 induces the late stage of osteoclastogenesis through DC-STAMP. In summary, as shown in the right part of Fig. 4., RANKL stimulates osteoclast precursors to form Trap-positive cells, which then produce CCN2; and this CCN2 induces the expression of DC-STAMP and interacts with the DC-STAMP to promote the fusion of the TRAP-positive cells, thus resulting in enhancement of late-stage osteoclastogenesis.

Moreover, CCN2 enhances the early stage of osteoclastogenesis (Aoyama et al. 2013). As shown in the left part of Fig. 4, we revealed by performing the solid-phase binding assay and using surface plasmon resonance (SPR) that CCN2 binds to RANK. SPR revealed that CCN2 binds to RANK with a KD value of less than 100 nM (Aoyama et al. 2013). Interestingly, CCN2 also binds to osteoprotegerin (OPG), which is a decoy receptor of RANK (Aoyama et al. 2013). Translocation of NF-κB into the nucleus is known to be one of the signaling pathways of RANKL (Hu et al. 2008); and CCN2 enhances this RANKL-induced nuclear translocation of NF-κB in RAW264.7 cells, which are osteoclast precursor cells (Aoyama et al. 2013). In addition, CCN2 enhances RANKL-induced phosphorylation of JNK, ERK, and p38 in RAW264.7 cells (Aoyama et al. 2013). Therefore, these findings strongly suggest that in the early stage CCN2 interacts with RANK and enhances RANKL signaling, leading to osteoclastogenesis (Aoyama et al. 2013). This effect may be due to increased availability of RANKL as a result of the binding of CCN2 to OPG. However, since we also found that OPG inhibits the interaction between CCN2 and RANK, OPG might inhibit the enhancement of RANKL signaling by CCN2 (Aoyama et al. 2013).

In conclusion, CCN2 also stimulates osteoclastogenesis in addition to acting on chondrocytes, osteoblasts, and endothelial cells. These findings strongly support our hypothesis that CCN2 is an endochondral ossification genetic factor (“ecogenin,” Fig. 5). However, this hypothesis comes from the data obtained mainly from in vitro experiments (Fig. 5). Therefore, we needed to confirm this hypothesis by performing in vivo studies, the results of which are described in the next section.

Fig. 5
figure 5

CCN2 enhances osteoclastogenesis in addition to promoting actions toward chondrocytes, osteoblasts, and vascular endothelial cells. In other words, CCN2 promotes endochondral ossification by acting on all four types of cells that are involved in endochondral ossification, thus providing the so-called “Ecogenin” (Endochondral ossification genetic factor) action of CCN2. Modified from Takigawa et al. (2003)

Full size image

Generation of transgenic mice overexpressing CCN2 in cartilage proves the “ecogenin” action of CCN2

To confirm that CCN2 actually promotes endochondral ossification in vivo, we generated transgenic mice over-expressing a ccn2/lacZ fusion gene in cartilage under the control of the 6 kb-Col2a1-enhancer/promoter (Tomita et al. 2013). Although there was no difference between WT and TG mice on embryonic day 15.5, the TG mice on postnatal day 1 were larger than the WT ones. Also at 8 weeks, these TG mice were larger than the WT animals (Tomita et al. 2013). When we measured the length of their long bones, the tibial length was longer in the TG mice than in the WT ones on postnatal day 1. Statistical analysis using P2 mice also confirmed that the TG bone was longer than the WT bone. Newborn transgenic mice also showed extended length of their long bones (Tomita et al. 2013). In order to assess whether the enhanced bone growth of CCN2 transgenic animals was due to enhanced cell proliferation, we stained sections of E19.5-day transgenic and wt embryos with an antibody against proliferative cell nuclear antigen (PCNA). The data showed that over-expression of CCN2 stimulated chondrocyte proliferation predominantly in the proliferative zone, but also in the resting zone (Tomita et al. 2013). Histological and immunohistological analyses revealed increased accumulation of proteoglycans and collagen II in the Tg mice (Tomita et al. 2013). The results of a TUNEL assay revealed marginally enhanced accumulation of apoptotic cells at the cartilage-bone interface and in the adjacent subchondral zone in the transgenic embryos as compared with that in the wild-type mice (Tomita et al. 2013). In in vitro short-term cultures of chondrocytes prepared from the cartilage of ccn2-over-expressing mice, the expression of col2a1, aggrecan and ccn2 was substantially enhanced; and in long-term cultures the expression levels of these genes were further enhanced (Tomita et al. 2013). The expression of Col10a1, a marker of hypertrophy, and that of vegf and of mmp-9, both vascular invasion factors expressed in the hypertrophic zone and boundary between cartilage and bone, were also enhanced; however, the level of stimulation in the transgenic chondrocytes was not as high as that for aggrecan or Col2a1 (Tomita et al. 2013). In addition, in vitro chondrogenesis by rib chondroblasts from ccn2-over-expressing mice was strongly enhanced (Tomita et al. 2013). IGF-I and IGF-II mRNA levels were elevated in the transgenic chondrocytes, and treatment of non-transgenic chondrocytes with CCN2 stimulated the expression of these mRNAs (Tomita et al. 2013). The addition of CCN2 induced phosphorylation of IGFR, and ccn2-overexpressing chondrocytes showed enhanced phosphorylation of IGFR, suggesting that our observation made on TG cartilage may have been mediated in part by CCN2-induced overexpression of IGF-I and IGF-II (Tomita et al. 2013). Micro-CT analysis revealed that overexpression of CCN2 increased bone density, the extent of mineralization of cancellous bone, and the thickness of the cortical bone (Tomita et al. 2013).

These findings indicate that overexpression of CCN2 in transgenic mice accelerated the endochondral ossification processes by promoting the proliferation and differentiation of growth-plate chondrocytes, resulting in increased length of their long bones (Tomita et al. 2013; Fig. 6). These results also indicate the possible involvement of locally enhanced IGF-I or IGF-II in this extended bone growth (Tomita et al. 2013).

Fig. 6
figure 6

Overexpression of CCN2 in the cartilage of transgenic mice promotes proliferation, differentiation, and hypertrophy of chondrocyes, resulting in elongation of the long bones. In other words, this in vivo experiment confirmed that CCN2 promotes endochondral ossification. The figure was drawn with help by Drs. N. Tomita and T. Hattori

Full size image

Chondroprotective action of CCN2 in articular cartilage shown by the transgenic mice overexpressing CCN2 in cartilage

CCN2 stimulates the synthesis of DNA and proteoglycans in articular chondrocytes as in growth-plate chondrocytes, but unlike its stimulation of calcification by growth cartilage cells, CCN2 does not stimulate undesired calcification in articular chondrocytes (Nishida et al. 2002; Kubota and Takigawa 2007b, 2011). Also, the administration of CCN2 together with gelatin hydrogel into cartilage defects repairs articular cartilage (Nishida et al. 2004; Kubota and Takigawa 2007b, 2011). These facts let us to speculate some special effect of CCN2 on the articular cartilage in these CCN2-overexpressing mice described above.

To investigate the effect of CCN2 overexpression on articular cartilage (Itoh et al. 2009, 2011), we used a spontaneously occurring osteoarthritis mouse model. We kept the WT and the TG mice for approximately 20 months and then carried out radiographic analysis of their knee joints. The results indicated that 50 % of the WT mice showed an osteoarthritis-like phenotype in their knee joints, as indicated by a narrowed joint space and rough cartilage surface (Itoh et al. 2009, 2011). On the other hand, no TG mice showed such an age-related osteoarthritis-like change (Itoh et al. 2009, 2011).

To confirm the enhanced expression of the ccn2 transgene in the articular cartilage in these old mice, we performed immunohistochemical analysis with ant-CCN2 antibody and found enhanced accumulation of CCN2 in the superficial and deep zones of the articular cartilage of knee joints from 21-month-old TG mice (Itoh et al. 2009, 2011). Also, histological analysis showed that TB staining, which is a marker of proteoglycan accumulation, was more intense in the TG mice than in the WT ones (Itoh et al. 2009, 2011). Moreover, immunostaining with anti-type II collagen showed that the collagen content did not decrease during aging in the TG mice, whereas it did decrease in the WT mice (Itoh et al. 2009, 2011). It is well known that collagen type I and X and MMP-13 are not detectable in normal articular cartilage and that the appearance of these proteins in articular cartilage is indicative of osteoarthritis (Kawaguchi 2008; Aigner et al. 1993; Shlopov et al. 2000). Therefore, we investigated the presence of these proteins in articular cartilage of these mice by immunostaining. As a result, age-related osteoarthritic changes such as the production of collagen types I and X and MMP-13 were not detected in the articular cartilage of the CCN2-overexpressing mice, in contrast to the positive staining for these three proteins in the WT articular cartilage (Itoh et al. 2009, 2011). In conclusion, cartilage-specific overexpression of CCN2 protected against the development of osteoarthritic changes in aging articular cartilage (Itoh et al. 2009, 2011). Thus, CCN2 may play a role as an anti-aging (chondroprotective) factor, stabilizing articular cartilage.

Molecular mechanism of CCN2-induced harmonized skeletal growth and regeneration

As was shown in Fig. 2, in the case of growth cartilage cells, CCN2 promotes not only proliferation and cartilage matrix formation, but also hypertrophy and calcification (Nakanishi et al. 2000). The latter are characteristics of growth cartilage cells. However, in the case of articular cartilage cells, CCN2 does not cause hypertrophy or calcification, which are undesired and pathological characteristics for articular cartilage (Nishida et al. 2002). Moreover, in the case of auricular chondrocytes, CCN2 promotes their proliferation and formation of cartilage matrix, which includes elastin, a major marker of elastic cartilage; but CCN2 does not cause hypertrophy of or calcification by these auricular chondrocytes (Fujisawa et al. 2008). Moreover, in vivo administration of CCN2 with gelatin hydrogel into a defect in articular cartilage regenerates the cartilage without undesired calcification (Nishida et al. 2004). Furthermore, CCN2 stimulates the expression of the differentiated phenotype of different types of cells such as osteoblasts (Nishida et al. 2000; Takigawa et al. 2003; Kawaki et al. 2008b, 2011), vascular endothelial cells (Shimo et al. 1998, 1999; Takigawa et al. 2003), and osteoclasts (Nishida et al. 2011a). CCN2 is also expressed in osteocytes (Kawaki et al. 2011) in developing bone; and mechanical stress, which is important for bone homeostasis, induces CCN2 expression in osteocytes (Yamashiro et al. 2001). CCN2 is also required for intramembranous ossification (Kawaki et al. 2008b), and in vivo adiministration of CCN2 with gelatin hydrogel into the defect of bone regenerates bone through intramembrabous ossification (Kikuchi et al. 2008). Moreover, CCN2 is expressed in sites of rat alveolar bone regeneration after tooth extraction (Kanyama et al. 2003) and during distraction osteogenesis (Kadota et al. 2004), suggesting that CCN2 plays an important role in regeneration. In other words, CCN2 promotes harmonized skeletal growth and maintenance.

How does CCN2 exhibit such a harmonizing action? What is the molecular mechanism of its action? CCN proteins have four modules, and many investigators including us have reported that each module has various binding partners, as described below. This might be a major reason for its harmonizing action (Fig. 3).

Extracellular matrices (ECMs) and integrins

Using 125I-labled CCN2, we previously revealed that CCN2 binds to perlecan, a heparan sulfate proteoglycan (Fig. 3, Nishida et al. 2003); but at that time we could not identify which domain was involved in this binding. Searching for CCN2-interactive proteins by yeast-two-hybrid screening, we identified the fibronectin 1 gene product as a major binding partner of CCN2 in the chondrocytic cell line HCS-2/8 (Fig. 3, Hoshijima et al. 2006). Only the CT domain of CCN2 bound directly to fibronectin (FN). CCN2 and its CT domain enhanced the adhesion of HCS-2/8 cells to FN. The CCN2-enhancing effect on cell adhesion to FN was abolished by anti-alpha5beta1 integrin (alpha5beta1), suggesting that CCN2 enhanced chondrocyte adhesion to FN through direct interaction of its C-terminal CT domain with FN and that integrin alpha5beta1 was involved in this adhesion (Hoshijima et al. 2006).

The yeast-two-hybrid screening also revealed that CCN2 binds to aggrecan and that its binding is through IGFBP and VWC modules (Fig. 3, Aoyama et al. 2009). SPR (surface plasmon resonance) analysis showed the direct interaction between CCN2 and aggrecan, and ectopically overexpressed CCN2 and G3 domain of aggrecan core protein confirmed their binding in vivo. When IGFBP-VWC modules, but not TSP-CT modules, were expressed in chondrocytic HCS-2/8 cells, aggrecan production in the cells was strongly induced, suggesting that this binding may be related to the CCN2-enhanced production and secretion of aggrecan by chondrocytes.

Moreover, CCN2 has been shown to bind to various ECMs such as heparan sulfate proteoglycan via its CT module (Nishida et al. 2003; Gao and Brigstock 2004), matrilin-3 via IGFBP-VWC modules (Araki et al. 2008), and decorin, in which case the binding module in the CCN2 molecule has not been identified (Vial et al. 2011; Fig. 3). CCN2 is also known to bind ECM-degrading enzymes such as matrix metalloproteinases (MMPs; Fig. 3), which cleave CCN2 (Hashimoto et al. 2002).

In addition to binding to integrin alpha5beta1, CCN2 and other CCN family members are reported to bind to various other integrins that are receptors for ECMs (for reviews, see Jun and Lau 2011; Lau and Lam 1999, and Kubota and Takigawa 2007b)

Growth factors and their receptors

Because it had been reported that CCN2 can antagonize BMP-4 activity by preventing its binding to BMP receptors in Xenopus embryos (Fig. 3, Abreu et al. 2002) and because both CCN2 (Kubota and Takigawa 2007b) and BMP-2 (Yoon and Lyons 2004) play an important role in cartilage metabolism, we first evaluated whether or not CCN2 could interact with BMP-2; and so we examined the combination effect of CCN2 with BMP-2 (CCN2-BMP-2) on the proliferation and differentiation of chondrocytes (Maeda et al. 2009). Immunoprecipitation-Western blotting analysis and a solid-phase binding assay revealed that CCN2 directly interacted with BMP-2, probably via its CT module and IGFBP-VWC modules (Fig. 3), and surface plasmon resonance (SPR) spectroscopy showed that its dissociation constant was 0.77 nM (Maeda et al. 2009). CCN2 was co-localized with BMP-2 at the pre-hypertrophic region in the E18.5 mouse growth plate (Maeda et al. 2009). Interestingly, CCN2-BMP-2 did not affect the BMP-2/CCN2-induced phosphorylation of p38 MAPK, but caused less phosphorylation of ERK1/2 in cultured chondrocytes (Maeda et al. 2009). CCN2-BMP-2 stimulated cell growth to a lesser degree than that achieved by either CCN2 or BMP-2 alone; whereas the expression of chondrocyte marker genes and proteoglycan synthesis, representing the mature chondrocytic phenotype, was increased collaboratively by CCN2-BMP-2 treatment in the cultured chondrocytes (Maeda et al. 2009). Because phosphorylation of ERK1/2 by CCN2 is involved in chondrocyte proliferation (Yosimichi et al. 2001, 2006), CCN2-BMP-2 treatment might promote chondrocyte differentiation by suppressing chondrocyte proliferation via decreased ERK1/2 phosphorylation. In any event, the above findings suggest that CCN2 may regulate the proliferating and differentiation of chondrocytes by forming a complex with BMP-2 as a novel modulator of BMP signaling (Maeda et al. 2009).

In addition to BMP signaling, FGF signaling is important in the developing endochondral skeleton (Ornits 2005). To investigate whether CCN2 interacts with fibroblast growth factor 2 (FGF2), we performed a solid-phase binding assay and immunoprecipitation-Western blot analysis and found that the C-terminal module of CCN2 (CT) directly bound to FGF2 (Fig. 3, Nishida et al. 2011b). SPR spectroscopy revealed that the dissociation constant between CCN2 and FGF2 was 5.5 nm. FGF2 promoted not only the proliferation of cultured chondrocytes but also the production of matrix metalloproteinase (MMP)-9 and −13 by them; however, FGF2 combined with the CT module, but not with full-length CCN, nullified the enhanced production of both MMPs and chondrocyte proliferation (Nishida et al. 2011b). CCN2 bound to the FGF receptor 1 with a dissociation constant of 362 nm, whereas the CT module did not (Fig. 3); and FGF2-stimulated phosphorylation of ERK1/2, p38 MAPK, and c-Jun N-terminal kinase was decreased by the CT module, but not by full-length CCN2 (Nishida et al. 2011b). These data suggested that the full-length CCN2 can activate intracellular FGF2 signaling, resulting stimulation of proliferation of chondrocytes and MMP production by chondrocytes. In contrast the cleaved CT module can bind to FGF2 but not to FGFR1 and so cannot activate FGF2 signaling (Nishida et al. 2011b). These findings suggest that CCN2 may regulate the proliferation and matrix degradation of chondrocytes by forming a complex with FGF2 as a novel modulator of FGF2 functions.

By screening a protein array, we also found that CCN2 could bind to fibroblast growth factor receptors (FGFRs) 2 and 3 (Aoyama et al. 2012). We ascertained that FGFR2 bound to CCN2 by performing a solid-phase binding assay and SPR. CCN2 enhanced the binding of FGFR2 to FGF2 and FGF4 at least through its TSP-1 module (Fig. 3). CCN2 and FGF2 had a collaborative effect on the phosphorylation of ERK and the differentiation of osteoblastic MC3T3-E1 cells. These results indicate the biological significance of the binding of CCN2 to FGFR2 in bone metabolism (Aoyama et al. 2012).

In addition to the growth factors and their receptors described above, CCN2 can bind to VEGF through TSP-1-CT modules (Inoki et al. 2002), to LRP-1 via the TSP-1 module (Segarini et al. 2001; Gao and Brigstock 2003), and to LRP-6, which is a Wnt co-receptor, via the CT module (Mercurio et al. 2004; Fig. 3). Concerning LRP-1, CCN2 was shown to induce tryrosine phosphorylation of its cytoplasmic domain in fibroblasts (Yang et al. 2004). However, in cartilage LRP-1 not only plays a role as a signaling receptor for CCN2 (Kawata et al. 2010) but also is involved in transcytosis of CCN2 protein in this avascular tissue, which is a protein transport system newly recognized in cartilage (Kawata et al. 2012).

Although its participating binding domain has not yet been identified, CCN2 is also known to interact with tropomyosin-related kinase A (Trk A), which is one of the neurotrophin receptors (Wahab et al. 2005). TrkA is observed in bone-forming cells during fracture healing (Asaumi et al. 2000) and distraction osteogenesis (Aiga et al. 2006) in both of which CCN2 has been shown to play an important role (Nakata et al. 2002; Kadota et al. 2004). CCN2 (Asano et al. 2005) and TrkA (Tsuboi et al. 2001) are also expressed in periodontal ligament cells, suggesting possible interaction therein.

Specific receptors

By means of a conventional receptor binding assay using iodinated CCN2 and a cross-linking study , a 280-kDa cell-surface molecule was identified on chondrocytes (Nishida et al. 1998), osteoblasts (Nishida et al. 2000), and vascular endothelial cells (Takigawa 2003). However, the structure of this molecule has not been clarified. Although we cannot exclude the possibility that this molecule is identical to one of the already known binding partners, cloning of this molecule is highly expected.

CCN2 proteins

Yeast two-hybrid screening using a cDNA library derived from a chondrocytic cell line, HCS-2/8, identified CCN2 and CCN polypeptides as CCN2-binding proteins (Hoshijima et al. 2012). Direct binding between CCN2 and CCN2 and between CCN2 and CCN3 was confirmed by co-immunoprecipitation and SPR, and the calculated dissociation constant (K d) was 1.17 × 10−9 M between CCN2 and CCN2, and 1.95 × 10−9 M between CCN2 and CCN3 (Hoshijima et al. 2012). Endogenous CCN2 and CCN3 interact and co-localize in chondrocytic HCS-2/8 cells, and ectopically overexpressed CCN2 and CCN3 also interact and co-localize in COS7 cells, as determined by direct fluorescence analysis (Hoshijima et al. 2012). CCN2-CCN3 interactions modulate CCN2 activity such as causing enhancement of aggrecan and col2a1 expression (Hoshijima et al. 2012). Curiously, CCN2 enhances, whereas CCN3 inhibits, the expression of aggrecan and col2a1 mRNAs in HCS-2/8 cells (Hoshijima et al. 2012) and proteoglycan synthesis by mouse rib chondrocytes in primary culture (Kawaki et al. 2008a); and the combined treatment with CCN2 and CCN3 abolishes the inhibitory effects of CCN3 (Hoshijima et al. 2012; Kawaki et al. 2008a). An antibody against the VWC domain of CCN2 (11H3) diminishes the binding between CCN2 and CCN2, but enhances that between CCN3 and CCN2 (Hoshijima et al. 2012). The results suggest that CCN2 can form homotypic and heterotypic dimers with CCN2 and CCN3, respectively. Enhanced binding between CCN2 and CCN3 with the 11H3 antibody has an enhancing effect on aggrecan expression in chondrocytes, suggesting that the CCN2-CCN3 heterodimer has a positive effect on aggrecan expression in chondrocytes (Hoshijima et al. 2012). We previously reported that CCN2 stimulates the gene expression of CCN2 itself and inhibits that of CCN3, that CCN3 inhibits the gene expression of CCN2 and has no effect on CCN3 expression, and that expression of CCN3 occurs earlier than that of CCN2 in the process of chondrocyte differentiation toward endochondral ossification (Kawaki et al. 2008a). Therefore, CCN3 suppresses CCN2 expression and chondrocyte differentiation in the early stage of chondrocyte differentiation, but after CCN2 starts to be expressed, the expressed CCN2 inhibits CCN3 expression and stimulates CCN2 expression, resulting in the promotion of differentiation. In this process, the CCN2-CCN3 heterodimer and then next the CCN2 homodimer would act as chondrocyte differentiation factors in this order.

In summary, CCN2 not only acts through its own receptors but also modifies the actions of various growth factors by binding to them and their receptors (Fig. 7). Moreover, it also binds to ECMs, which accumulate growth factors around the cells and modifies the actions of these growth factors (Fig. 7). The sum of these actions would become the final outcome of biological action of CCN2 (Fig. 7). Moreover, gene expression and activity of CCN2 is also regulated by another member of the CCN family of proteins, CCN3 (Fig. 7). In other words, various biological actions of CCN2 would depend on the different microenvironments of various types of cells as well as on their state of differentiation. Through these various mechanisms, CCN2 would orchestrate various signaling pathways by acting as a signal conductor to promote harmonized skeletal growth and maintenance (Fig. 7).

Fig. 7
figure 7

How CCN2 exhibits such a multifunctionality? CCN2 not only acts through its own receptors but also modifies the actions of various growth factors by binding to them and their receptors. Moreover, it also binds to ECM, which accumulates growth factors around the cells and modifies the actions of these growth factors. The sum of these actions would become the final outcome of the biological action of CCN2. In other words, the various biological actions of CCN2 would depend on the different microenvironments of various types of cells and the state of their differentiation

Full size image

Future prospects and conclusions

Evidence thus far indicates that the most important physiological role of CCN2 is the promotion of endochondral ossification. Roles in regeneration, protection of bone and articular cartilage, and angiogenesis are also important ones for CCN2. Continuous search efforts to find other binding partners of CCN2 and to identify the corresponding binding modules in the CCN2 molecule, as well as further investigation of combinational effects, may lead to a better understanding of the molecular mechanisms responsible for the multifunctionality of the CCN2 protein.