2-D08 br Discussion The N terminal domain of DDRs
Discussion The N-terminal domain of DDRs has long been recognized as a member of the DS superfamily (Johnson et al., 1993, Karn et al., 1993), and its role in collagen binding is understood in atomic detail (Carafoli et al., 2009, Ichikawa et al., 2007). Our crystal structure shows that the second DDR domain is a distant relative of the DS domain, termed the DS-like domain. Tandem repeats of DS domains occur in a number of secreted and cell surface 2-D08 (Baumgartner et al., 1998, Kiedzierska et al., 2007). In the blood coagulation factors V and VIII, the two DS domains are arranged side by side with limited contacts between them, so that their top loops can both interact with the same cell membrane (Adams et al., 2004, Ngo et al., 2008, Shen et al., 2008). In neuropilin-1 and -2, the two DS domains are related by a ∼90° rotation and form a compact structure, as in DDR1 (Appleton et al., 2007, Vander Kooi et al., 2007). This angled arrangement in DDR1 results in the C terminus of the DS-like domain emerging near the interdomain linker. The presumably unstructured JM region of DDR1 linking the DS-like domain to the TM helix (residues 368–417) contains 12 prolines and a number of predicted N- and O-linked glycosylation sites. If fully extended, it would project the DS and DS-like domains of DDR1 ∼150 Å from the cell surface. The JM regions of other DDRs are similarly long, ranging from 32 to 74 residues. mAbs directed against RTKs are invaluable tools for research and have been developed into successful therapeutics (Adams and Weiner, 2005). We have characterized seven anti-DDR1 mAbs that inhibit DDR1 function by binding to two distinct regions in the DS-like domain. Notably, Fab fragments derived from these mAbs were equally effective as DDR1 inhibitors. No mAbs were obtained that bind to the DS domain, possibly reflecting the higher degree of surface conservation in that domain (not shown). In agreement with their epitope locations, the mAbs inhibit DDR1 function without blocking collagen binding. We think that they do so by preventing the proximity of the two DS-like domains and the following JM regions in the collagen-bound, signaling, state of the DDR1 dimer (Noordeen et al., 2006). Deletion of the DS-like domain or JM region of DDR1 results in receptors that are not trafficked to the cell membrane, so the contribution of these regions to signaling could not be studied (Noordeen et al., 2006). Remarkably, however, the DS-like domain of DDR2 could be deleted without abrogating collagen-induced receptor autophosphorylation (Leitinger, 2003), suggesting that the DS-like domain is not making any essential contacts in the signaling DDR dimer. This leaves the collagen-bound DS domain as the most likely site of contact between the extracellular regions of the two DDR protomers in the signaling dimer. An analysis of crystal lattice contacts in the DDR1-3E3 Fab structure led to the fortuitous discovery of functionally important residues near the base of the DS domain, close to the interface with the DS-like domain and distant from the collagen-binding site at the top of the DS domain. The patch formed by these residues is the largest concentration of conserved surface residues in the extracellular region of DDRs apart from the collagen-binding site, consistent with its essential role in signaling. We think that the conserved patch is involved in mediating protomer contacts in the signaling DDR dimer, either by forming a direct DS-DS interface or by providing a secondary collagen-binding site. The latter alternative is more appealing, because it provides a plausible mechanism whereby collagen could cross-link two DS domains (analogous to the “composite binding site” model discussed by Carafoli et al., 2009). In solution, the isolated DDR2 DS domain binds a 28-residue collagen peptide with 1:1 stoichiometry (Carafoli et al., 2009). However, inspection of the crystal lattice of this DS-collagen complex reveals that the conserved patch is involved in a lattice contact with the N-terminal glycine-proline-hydroxyproline triplets of the collagen peptide. This intriguing observation may suggest that the conserved patch in the DS domain indeed has weak affinity for collagenous sequences and, therefore, could provide a secondary collagen-binding site in dimeric, full-length DDR.