In order to further understand

In order to further understand the biological significance of cell-collagen X interactions, we have examined the possibilities of involvement of other collagen receptors in cell binding to collagen X. The present study demonstrates that collagen X is a ligand for DDR2. DDR2 has been observed to bind to a number of fibrillar collagens (types I, II, III and V), but does not recognise the basement membrane collagen IV (Shrivastava et al., 1997, Vogel et al., 1997). Collagen X thus represents the first non-fibrillar ligand for DDR2. The solid phase binding assays (Fig. 2) and receptor autophosphorylation data (Fig. 3) show that collagen X is primarily a ligand for DDR2, not DDR1. However, as DDR1 showed weak binding to collagen X, DDR1 may act as a low affinity collagen X receptor. These results are similar to our earlier data on the interaction of the DDRs with the cartilage-specific collagen II (Leitinger et al., 2004). For a physiologically relevant interaction, ligand and receptor have to be expressed in the same place. Collagen X LLY507 is restricted to the growth plate of long bones, while DDR2 is expressed widely in a number of tissues throughout the body. Our data presented in Fig. 4, Fig. 5 show that hypertrophic chondrocytes, the source of collagen X deposition, express mRNA for DDR2 and that the DDR2 protein can also be detected in hypertrophic chondrocytes. These findings are consistent with an interaction of DDR2 with collagen X in the growth plate and strongly support the idea that DDR2 is a physiological receptor for collagen X. In an earlier study, DDR2 mRNA was detected in proliferating murine chondrocytes in vivo, and the elimination of DDR2 in the mouse led to a bone growth defect due to reduced chondrocyte proliferation (Labrador et al., 2001). Our present RT-PCR data are consistent with the report by Labrador et al. (2001). However, we have not been able to localise DDR2 protein in the proliferation zone of the murine growth plate by immunohistochemical methods. Fig. 5F indicates that the initial DDR2 protein expression is at the junction of the proliferative and hypertrophic zones. Our observations in the growth plate therefore suggest a possible role of DDR2 in cell maturation rather than cell proliferation. In the articular cartilage, DDR2 protein is detected on all articular chondrocytes whereas collagen X expression is restricted to the pericellular area of the deep zone (hypertrophic) chondrocytes (Fig. 5A and B). The reason why DDR2 is not found on growth plate proliferative chondrocytes would require further investigation. Thus, DDR2 fulfils important functions in bone growth, and our present study forms the basis for future studies into the effect of DDR2 signalling on hypertrophic chondrocytes and EO. In particular, our hypothesis that DDR2 regulates chondrocyte maturation will need to be tested. Our previous studies have demonstrated that the DDR2 discoidin domain fully contains the binding site(s) for the fibrillar collagens I and II (Leitinger, 2003, Leitinger et al., 2004). The binding site for collagen I was mapped to three spatially adjacent surface loops within the DDR2 discoidin domain (Leitinger, 2003). To our surprise, collagen X was not recognised by the DDR2 discoidin domain (Fig. 6), indicating additional requirements for collagen X recognition. From these data it seems that the binding mechanism for non-fibrillar collagens (or binding sites on DDR2) is different from how fibrillar collagens are recognised by DDR2. At present, we cannot rule out that the collagen X binding site lies outside the discoidin domain, in the region between the discoidin domain and the transmembrane domain. This ∼200 amino acid sequence shows no sequence homology to any known domain. However, collagen recognition by the DDRs most likely occurs via a conserved mechanism and the discoidin domain is situated at the membrane-distal site of DDR2, where one would expect a collagen binding site to be located. We therefore speculate that the DDR2 discoidin domain contains at least part of the collagen X binding site, but that additional protein sequences are required for maintaining the proper orientation of the discoidin domain for collagen X recognition. We have previously shown that DDR binding to collagen requires the extracellular DDR domains to be dimerised (Leitinger, 2003). It is conceivable that the DDR2 discoidin domain construct (dimerised via an Fc sequence) does not present the discoidin domain in the correct orientation for collagen X binding.