For a physiologically relevant interaction
For a physiologically relevant interaction, ligand and receptor have to be expressed in the same place. Collagen II is found primarily in cartilage. DDR2 is expressed in proliferating chondrocytes in vivo, and its staining pattern in tibial growth plates is similar to that of collagen II. In addition, the elimination of DDR2 leads to shortening of long bones due to impaired chondrocyte proliferation. Moreover, apoptosis of chondrocytes, and alteration of cartilage and some elements of the vertebrate skeleton were demonstrated in mice with inactivated Col2α1 gene. These findings are consistent with an interaction of DDR2 with collagen II in bone and support the idea that DDR2 is a physiological receptor for collagen II. To our knowledge, no data are available about the expression of DDR1 in cartilage. In DDR1 −/− mice, chondrocyte proliferation and apoptosis is not affected. Here, we demonstrate the presence of a specific DDR2 binding site in the D2 period of collagen II, located between amino G-1 residues 235 and 468 of the collagen triple helix. The DDR-collagen interaction is strictly dependent on the triple-helical conformation of collagen. Our approach to localise the DDR2 binding site of collagen II was therefore based on the generation, through genetic engineering, of well-defined triple-helical collagen II proteins that cover the entire amino acid sequence of the monomer. We favoured this approach over a more classical one that would use cyanogen bromide-derived collagen peptides, as these collagen fragments are not triple-helical at physiological temperatures, and previous attempts by other researchers to purify all peptides for collagen II failed. Elimination of the D2 period from collagen II led to a complete loss of DDR2 binding activity (Figure 5), suggesting that the D2 period contains the major, high-affinity DDR2 binding site in collagen II. Deletion of D1 from collagen II resulted in a partial loss of DDR2 binding (Figure 5). This could be explained by the presence of an auxiliary DDR2 binding site in D1, and we cannot formally exclude this possibility. However, the D1 period is required for proper folding of the N-terminal propeptide, and its deletion affects the stability of the triple helix negatively. While the other three D domain-deleted collagen II variants show melting curves similar to those of the full-length protein, for −D1 procollagen II the temperature for 25%, 50% and 75% unfolding was 2–4°C lower. The lack of the D1 period likely results in significant conformational changes, as evidenced by a non-specific cleavage of the −D1 procollagen II with procollagen N-proteinase. These findings make it very likely, in our opinion, that the reduced DDR2 binding of the −D1 construct is due to a structural effect rather than a loss of a specific binding sequence. Significantly, the presence in the multi-D4 construct of a single D1 domain, included for its requirement for proper folding, did not result in DDR2 binding or enhanced autophosphorylation (Figure 8, Figure 9), indicating that D1 alone, without the presence of D2, is not sufficient for DDR2 recognition. Therefore, we believe that D1 does not contain a critical DDR2 binding site. The deletion of D1 from collagen II resulted in a complete loss of DDR2 autophosphorylation (Figure 6). As the cellular assay is performed at 37°C, in contrast to the solid-phase binding assays performed at room temperature, a substantial proportion of the −D1 construct will unfold and thereby lose its triple-helical conformation at 37°C, which could account for the effect on DDR2 activation. Collagen-induced receptor phosphorylation might also have more stringent requirements than the in vitro binding of recombinant DDR ECDs to collagen. As the melting curves of the −D2 and the multi-D4 constructs overlap completely with that of the full-length, normal procollagen II,15, 17 the loss of autophosphorylation in response to these collagen II variants cannot be explained by an effect on the triple-helical conformation. Interestingly, the multi-D2 protein, which begins to unfold at a temperature 3°C lower than multi-D4 and normal procollagen II, induced stronger DDR2 autophosphorylation than the normal procollagen II (Figure 9). However, the effect was not as great as might be expected if the three D2 binding sites acted in an independent, additive manner. It is likely that DDR2 on the surface of cells cannot interact simultaneously with all three binding sites on the multi-D2 molecule.