Three DDR binding sites have been mapped on the collagen
Three DDR2 binding sites have been mapped on the collagen triple helix by us for collagen type 1 and by others using the collagen toolkit for collagen type 2. All of the three reported binding sequences are conserved in the α1 chain of collagen types 1, 2 and 3. The central motif sequence GARGQAGVMGFO corresponding to methylguanine dna methyltransferase 394–405 has the highest binding affinity for DDR2. This binding site overlaps with the binding site of another soluble collagen-binding protein von Willebrand factor (vWF) in collagen type 3, and is in close proximity to a binding site for decorin on collagen type 1. Although no study has been reported on the effect of vWF on collagen fibrillogenesis, decorin has been found to result in smaller collagen fiber diameter,2, 29 similar to that observed by us for DDR2. It is interesting to note that majority of the collagen-binding proteins known to modulate collagen fibrillogenesis are glycoproteins. Glycosylation of collagen is considered an important factor for binding of both decorin and DDR27, 8 to collagen. Additionally, collagen glycosylation on the hydroxylysines can have a critical role in regulating fiber diameter in collagen fibrillogenesis.31, 32 Studies addressing the role of glycosylation in DDR2 binding to collagen need further exploration, especially since DDR2 is a glycoprotein with at least one of the binding sites in close proximity to decorin. Changes in collagen fiber diameter and rate of deposition can also arise due to various types of collagens (heterologous collagen) being incorporated into a fiber. In tissue cultures, Contard et al. showed that fibers with measured diameters of 34 nm or less bound antibodies against collagen type I molecules, while antibodies against collagen type III bound fibers with a diameter between 35 nm and 54 nm. In vitro studies by Birk et al. demonstrated that collagen fiber diameter is regulated by collagen type V. They found that pure collagen type V forms small fibers (mean 25 ± 8 nm) with no apparent banded structure seen by TEM. Type V and type I collagens are often known to form mixed fibers with the average collagen diameter increasing with percentage of type I collagen. The collagen fibers assembled in the ECM of mouse osteoblast cells used in this study are composed mostly of collagen types 1, 3 and 5. Further studies are required to determine if DDR2 has varying affinities for these different collagen types and if expression of DDR2 ECD results in differences in collagen composition of the fibers formed. Nevertheless, since our earlier in-vitro results with purified collagen type 1 and purified DDR2 ECD also showed the effects of DDR2 on collagen fibrillogenesis, it is unlikely that our observed differences in the present study are largely due to differences in the expression levels of different collagen types. DDR2 knock-out mice have been shown to possess skeletal defects such as shortening of long bones and irregular growth of flat bones. These defects in knock-out animals have been explained on the basis of impaired chondrocyte and fibroblast proliferation observed in the absence of DDR2. While it is well-known that the ECM can influence cell proliferation, no report exists so far on the ultrastructural collagen morphology for DDR2 knock-out animals. Our results indicate that expression of DDR2 may be critical to regulate collagen deposition, which in turn may affect cell proliferation. A detailed examination of the ECM morphology in DDR2 knock out versus wild type animals will provide a more complete understanding of the role of DDR2 in matrix turnover and cell proliferation. The collagen receptor DDR2 (and likely DDR1) can regulate collagen by two mechanisms: by activating and upregulating MMPs, as reported earlier, and by inhibition of collagen fibrillogenesis as demonstrated in our studies. These two mechanisms give rise to a weakening of the ECM that can influence cell adhesion, migration and proliferation. One may speculate that a weakened ECM would play a different role in developing versus adult tissues. In adult tissue, a weakened ECM could result in heightened tumor invasiveness, which is compatible with findings of DDR2 over-expression in malignancies.9, 10, 11, 12 In developing tissue, it is possible that a weaker, or more dynamic, ECM is needed for cell proliferation; such has been reported for the developing heart.