It is well established that Shp can function as

It is well established that Shp-2 can function as a substrate for several RTK, such as PDGF or EGFR [23], [36]. To test whether DDR1 recognizes Shp-2 as its substrate, we overexpressed a catalytically inactive Shp-2 mutant together with DDR1b in 293 cells [31]. Immunoprecipitation of Shp-2 followed by anti-phosphotyrosine Western blotting revealed a DDR1 activation-dependent phosphorylation of catalytically inactive Shp-2 (Fig. 4B). Notably, we neither detected phosphorylation of the catalytically inactive mutant in the absence of co-transfected DDR1 (data not shown), nor of wild type Shp-2 in the experiments displayed in Fig. 3B, which suggests that Shp-2 regulates its own state of phosphorylation. Lastly, we confirmed the interaction of DDR1 with endogenous Shp-2 in MCF-7 cells and found that higher amounts of Shp-2 bound to collagen-activated DDR1, than to DDR1 in unstimulated cells (Fig. 4C). Our previous work showed that mice lacking DDR1 are viable and healthy but fail to lactate at parturition [40]. We reasoned therefore that the mammary gland epithelium is the most appropriate tissue to further define the functional relevance of the interaction between DDR1 and Nck2 or Shp-2. To this end, mammary gland tissue sections from wild type and DDR1-null mice after the onset of lactation (2 days post partum, 2dpp) were stained with Senexin A for Nck2 or Shp-2. In wild type tissue, highest Nck2 expression was found in focal membrane-proximal areas of luminal epithelial cells (Fig. 5A, indicated by arrows). In contrast, Nck2 was diffusely expressed throughout the cytoplasm in DDR1-null tissue (Fig. 5B). This aberrant staining was specific for tissue from lactating DDR1-null mice, since during pregnancy we observed staining undistinguishable to wild type samples (data not shown). Staining for Shp-2 clearly confirmed its cytoplasmic localization, albeit no difference between wild type and DDR1 knockout was apparent (Fig. 5 C and D). The aberrant Nck2 distribution is most likely not caused by cytoskeletal differences between knockout and wild type, since staining for F-actin is similar in both cases (Fig. 5 E and F). In summary, the current set of data places DDR1 upstream of at least two phosphotyrosine-dependent interactions mediated by the SH2 domains of Nck2 or Shp-2. In extension of earlier work, we show that the Nck2 and Shp-2 are critical regulators for cell differentiation of mammary gland cells [10], [15].
Discussion Full-length DDR1 (c-isoform) has a total of 15 tyrosines in its cytoplasmic region and a phosphopeptide mapping approach performed here suggests that several of these tyrosines are phosphorylated upon DDR1 activation (Fig. 3A). The amino acid sequence pY-D-E-P had been identified as a consensus binding site for the SH2 domain of Nck1, which has a 68% overall identity to Nck2 [5], [30]. However, no clear distinction between the binding preferences of Nck1 and Nck2 has yet been made. Within the DDR1 cytoplasmic tail, tyrosine-484 (Y-Q-E-P) closely matches the Nck consensus, however further experimental work is required to confirm it as an Nck2 binding site. Biochemical and genetic evidence supports a central role for the two Nck adapters in cell migration [5]. The generation of knockout mice revealed that Nck1 and Nck2 have largely redundant functions as neither of the single knockouts resulted in a phenotypic alteration, while the Nck1/Nck2 double knockout was lethal during early embryogenesis [4]. Both proteins can form a wide array of interactions and serve as a link between cell surface receptors through its SH2 domain and a variety of effectors using their SH3 domains. Membrane recruitment of effectors such as Sam68, Pak or WASP trigger cytoskeletal remodeling, which results in increased chemotaxis, migration, axon extension or invasion [19]. Data analyzing DDR1 function in the mammary gland are in support of a pro-migratory function of Nck2 downstream of DDR1 because knockout mice show dramatically reduced epithelial duct elongation and branching at puberty [9], [40]. During late pregnancy and at parturition, DDR1 potentially places Nck2 to specific locations within the mammary epithelium, which may be essential for proper alveologenesis and for maintenance of lactation (Fig. 5). In leukocytes and glioma cells, overexpression of DDR1 results in enhanced cellular migration and invasion [17], [27]. Interestingly, Nck2, but not Nck1, interacts with focal adhesion kinase as well as with PINCH, a molecule associated with integrin-linked kinase, during integrin-mediated cell adhesion [10], [34]. It is tempting to speculate that integrin and DDR1 signaling pathways intersect by utilizing Nck2 as common adaptor molecule. This assumption is supported by the fact that in the developing nervous system, a specific role of Nck2 downstream of the TrkB tyrosine kinase and B-type ephrins have been suggested where activation of the Nck2 pathway results in growth cone retraction [7], [32]. Although we have only started to address the specific role of Nck2 downstream of DDR1, cell-wounding experiments with primary embryonic fibroblasts showed that cells lacking DDR1 are severely impaired in migration and wound closure (D.H.H.K., Y.H. and W.F.V., manuscript in preparation).