We can think of two ways in which
We can think of two ways in which collagen binding could activate DDR (Figure 8). A single collagen triple helix could interact with both DS domains in the DDR dimer (“composite binding site”) and thereby activate the receptor, similar to the situation exemplified by the growth hormone-growth hormone receptor complex (de Vos et al., 1992). The key collagen residues in our crystal structure are Nle21 and Hyp24 of the leading chain and Phe23 of the middle chain. Because of the helical symmetry of the homotrimeric collagen peptide, an equivalent constellation of residues occurs again on the middle and trailing chains (Figure 1B). However, it is impossible to replicate the interactions of the first DS domain at this second, vacant, site without causing major steric clashes between the two DS domains (data not shown). Thus, the complex would have to be asymmetric with two distinct receptor-ligand interfaces. The high-affinity DS-collagen interface would correspond to the interaction seen in our crystal structure, whereas the second interface may be weaker and only form when the two DS domains are joined in a stable DDR dimer.
In the alternative scenario, collagen binding to two independent sites would trigger the transition from the inactive to the active DDR dimer, conceivably by amplifying the small changes in the collagen-binding loops of the DS domain (Figure 3). This situation would be more akin to the EGF-EGF receptor complex, in which two EGF molecules bind to equivalent sites on the outside of an active receptor dimer (Burgess et al., 2003). It should be noted that there is no formal requirement for both DS domains to be occupied by ligand in the active DDR complex: binding of collagen to one DDR DS domain may be sufficient to “unlock” the inactive dimer. Structures of the full-length receptor will now be required to gain further insight into the mechanism of DDR activation. Our structure of a DDR2 DS-collagen complex provides the foundation for such future studies.
Introduction During cdc42 pathway (EO), chondrocytes within the cartilage template undergo a highly co-ordinated sequence of proliferation, maturation and hypertrophy with the concomitant changes in the synthesis and deposition of extracellular matrix (ECM) components at each stage (Ortega et al., 2004). Many paracrine and endocrine factors are now known to be regulators of EO (Hering, 1999, de Crombrugghe et al., 2000, van der Eerden et al., 2003). Indian Hedgehog and parathyroid hormone related protein have been shown to regulate chondrocyte hypertrophy in a negative feedback control loop (Vortkamp et al., 1996), possibly acting in conjunction with several other regulatory factors including bone morphogenetic proteins and fibroblast growth factors (Minina et al., 2001, Minina et al., 2002). The ECM also regulates EO by mediating cell migration and shape changes, cell proliferation and differentiation either through direct cell–matrix interactions or by facilitating growth factor binding to cells (Shimazu et al., 1996, Bass and Humphries, 2002). It has been generally accepted that the hypertrophic ECM serves as a permissive matrix to vascularisation and mineralisation. It is therefore conceivable that collagen X, which exhibits unique temporal and spatial expression patterns in the hypertrophic zone, would have specific regulatory roles in EO besides the maintenance of tissue structure and integrity. Collagen X is a homotrimeric molecule of three α1(X) chains (Mr 59 kDa) comprising a 45 kDa triple-helical domain flanked by an N-terminal (NC2) and a larger C-terminal (NC1) non-collagenous domains (Shen, 2005). In the hypertrophic ECM, collagen X most likely forms an extended hexagonal network, as shown by in vitro studies (Kwan et al., 1991) and by electron microscopy on the murine growth plate (Jacenko et al., 2001), and is particularly abundant in the pericellular matrix of hypertrophic chondrocytes (LuValle et al., 1992, Tselepis et al., 1996). Mutations of the human Col10A1 gene are known to cause Schmid metaphyseal chondrodysplasia, an autosomal dominant disorder characterised by short stature, widened growth plates, bowing of the long bones and coxa vara (Warman et al., 1993, Wallis et al., 1994, Mäkitie et al., 2005) with the majority of mutations being found in the NC1 domain (Chan and Jacenko, 1998). Skeletal defects characteristic of spondylometaepiphyseal chondrodysplasia were reported in mice expressing a truncated collagen X transgene containing a large in-frame deletion (Jacenko et al., 1993). These studies indicate that either a reduction collagen X deposition due to haploinsufficiency or disruption of the normal collagen X network due to dominant interference can lead to aberrant EO. Furthermore, the complete lack of collagen X deposition in the matrix of Col10A1 −/− mice resulted in growth plate compression, displacement of proteoglycans, altered mineral deposition, and hematopoietic changes (Kwan et al., 1997, Gress and Jacenko, 2000, Jacenko et al., 2002). Based on the disease phenotypes observed in these transgenic mouse models, the pericellular collagen X network appears to be an important link between the hypertrophic chondrocytes and the interterritorial matrix especially in stabilising the proteoglycan network. We therefore propose that the interactions between hypertrophic chondrocytes and the collagen X network are important in maintaining the integrity of the hypertrophic matrix and regulate chondrocyte metabolism through cell adhesion molecules. This hypothesis is partially supported by our recent findings that hypertrophic chondrocytes can adhere and spread on a collagen X substrate (Luckman et al., 2003). Hypertrophic chondrocyte adhesion to collagen X is primarily mediated through the α2β1 integrin. In this study, we report the interactions between collagen X and a non-integrin collagen receptor, the discoidin domain receptor DDR2.