The collagen receptor DDR1 regulates cell spreading and motility by associating with myosin IIA

The maintenance of tissue architecture requires that mammalian cells continuously and appropriately respond to extracellular matrix signals and mechanical stimuli (Geeves et al., 2005; Wakatsuki et al., 2003). These responses include the generation of propulsive and contractile forces, generated primarily by actin assembly and myosin motor activity (Dobereiner et al., 2005; Richards and Cavalier-Smith, 2005; Yamaguchi and Condeelis, 2007). Fifteen classes of myosin have been identified, all of which have a motor domain (Richards and Cavalier-Smith, 2005). The motor domain interacts with polymerized actin, hydrolyzes ATP and enables cellular movement. Non-muscle myosin IIA (NMHC-IIA; also known as MYH9) is a hexameric enzyme composed of two heavy chains, each with a regulatory chain and an essential light chain. The ATPase function of NMHC-IIA allows transient binding to the actin cytoskeleton, thereby generating mechanical forces that can maintain cellular architecture or initiate cell movement. In conjunction with actin filaments and microtubules, NMHC-IIA powers cell adhesion and migration in a diverse range of cultured cells (Even-Ram et al., 2007; Giannone et al., 2007; Vicente-Manzanares et al., 2007). Mouse embryos lacking NMHC-IIA die at day 6.5-7 of development, highlighting a fundamental role for this particular myosin in cellular function (Conti et al., 2004). Autosomal dominant mutations of NMHC-IIA in patients manifest as May-Hegglin syndrome, a disorder characterized by an array of symptoms including abnormal maturation of platelets and leukocytes, hearing loss, lens defects and renal failure (Pecci et al., 2005).

Discoidin domain receptor 1 (DDR1) is an unusual receptor tyrosine kinase as it responds to extracellular matrix components such as fibrillar collagens, but not to soluble growth factors (Shrivastava et al., 1997; Vogel et al., 1997). DDR1 is over-expressed in several human cancers and is a direct transcriptional target of the p53 tumor suppressor gene, highlighting a possible role in cellular transformation (Alves et al., 1995; Ongusaha et al., 2003). The kinetics of DDR1 activation are slow and sustained compared with other receptor tyrosine kinases, suggesting that this receptor may be involved in mediating longer-term signals in non-transformed cells (Faraci-Orf et al., 2006). DDR1 phosphorylation is stimulated by fibrillar and basement membrane collagens (currently, types I-VI and type VIII are known to activate), implying broad importance for cell adhesion and migration in various tissues. DDR1 can regulate migration of leukocytes and kidney epithelial cells in collagen-rich microenvironments in vitro (Kamohara et al., 2001; Wang et al., 2005). The adaptor molecules ShcA, Nck2 and Shp-2 bind to activated DDR1 in a phosphotyrosine-dependent manner (Kamohara et al., 2001; Koo et al., 2006; Vogel et al., 1997). The interaction of DDR1 with Nck2 is mediated by the SH2 domain of Nck2, and this interaction is implicated in triggering DDR1-induced downstream responses (Koo et al., 2006). In macrophages and T-cells, collagen-induced DDR1 activation mediates activation of the p38 MAPK and NFκB pathways (Kim et al., 2007; Matsuyama et al., 2004; Yoshimura et al., 2005). Although activation of DDR1 by collagen is independent of β1 integrin collagen receptors, activated DDR1 might inhibit α2β1-integrin-dependent signaling through the Stat1-Stat3 complex (Vogel et al., 2000; Wang et al., 2006). Immunohistochemistry has shown that DDR1 localizes to basement membrane contacts but it is not part of the focal adhesion complex (Sakamoto et al., 2001; Vogel et al., 2000) and it has not been shown so far to interact with cytoskeletal motor proteins.

The cellular functions of DDR1 identified from cell culture assays are supported by in vivo data. For example, DDR1-knockout mice show reduced post-natal skeletal growth (Alves et al., 2001) and adult females are unable to nourish their litters because the mammary gland epithelium fails to secrete milk (Alves et al., 2001; Faraci-Orf et al., 2006). Vascular smooth muscle and mesangial cells derived from knockout mice confirm that DDR1 is of critical importance for cell migration (Curat and Vogel, 2002; Hou et al., 2001). However, the mechanisms involved in triggering cell motility downstream of activated DDR1 are not understood. In the work reported here have examined the association and functional outcomes involving DDR1 and non-muscle myosin IIA after collagen stimulation.

We examined the protein expression levels of DDR1, NMHC-IIA and β1 integrin in the principal cell types that were used as in vitro models for the studies described below. There were high levels of DDR1 expression in MCF-7 cells and in NIH3T3 cells stably transfected with DDR1 (Fig. 1A). By contrast, DDR1 was not detected in NIH3T3 cells transfected with a control plasmid (pLXSN) and was present at very low levels in fibroblasts from DDR1 null mice. Notably, the levels of β1 integrin were relatively similar for all of the cell types except for the MCF-7 cells, which exhibited very low levels of β1 integrin. We confirmed that the NIH3T3 cells stably transfected with DDR1 (3T3-DDR1) were ligand-responsive by plating overnight on 10 μg/ml type I collagen and immunoblotting for tyrosine phosphorylation of DDR1 (Fig. 1B).

We next identified proteins associating with DDR1. Collagen-stimulated cells were lysed, DDR1 was immunoprecipitated and co-precipitating bands were analyzed by mass spectrometry. A DDR1-binding protein of approximately 200 kDa was detected by gel electrophoresis; mass spectrometry of tryptic peptides of this protein identified it as NMHC-IIA (Table 1). The relative amount of NMHC-IIA that coimmunoprecipitated with DDR1 was increased after 8 hours of plating on collagen, suggesting that the association between NMHC-IIA and DDR1 depends upon collagen-induced DDR1 activation (Fig. 2A). When similar procedures were applied to NIH3T3 cells, there was no detectable 200 kDa band (Fig. 2A, lower gel). We confirmed a DDR1–NMHC-IIA association by reciprocal coimmunoprecipitation with an antibody directed against NMHC-IIA, followed by detection with an anti-DDR1 antibody (Fig. 2B). Immunoprecipitations conducted with pre-immune serum showed no detectable DDR1 or NMHC-IIA. An association between NMHC-IIA and DDR1 was verified in human breast cancer cells (MDA-MB-231) that lack endogenous DDR1 after transfection with full-length DDR1b, but not with the truncated DDR1d isoform lacking the kinase domain (Fig. 2C). These cells exhibited strong expression of DDR1 when transfected with plasmids, which rationalized their use in these experiments. Knockdown of endogenous DDR1 with siRNA in MCF-7 cells eliminated NMHC-IIA from the DDR1 immunoprecipitates, supporting the specificity of this association (Fig. 2D). Immunoprecipitations prepared using antibody to NMHC-IIA effectively pulled down DDR1 in control siRNA-treated MCF-7 cells but not in siRNA knockdown cells (Fig. 2D, middle blots).

Although the association between DDR1 and NMHC-IIA was evidently stimulated by plating cells on collagen, the association between these two proteins occurred independently of collagen-binding integrin receptors since coimmunoprecipitation of endogenously expressed proteins isolated from integrin β1-deficient GD25 cellular extracts yielded results similar to those from wild-type fibroblasts (Fig. 2E). The strength of the association between DDR1 and NMHC-IIA was relatively high as the complex could be recovered either in buffer containing 1% Triton X-100 or in immunoprecipitation buffer, which contains more stringent detergents (data not shown). These more stringent buffer conditions also ruled out the possibility that coimmunoprecipitation of NHHC-IIA with DDR1 was caused by the high overall protein-protein binding affinity of cytoskeletal molecules. Furthermore, we did not find actin in the NMHC-IIA–DDR1 immunocomplexes.

We examined whether DDR1 associates with a specific domain of NMHC-IIA. The C-terminal rod domains from NMHC-IIA or -IIB were expressed as His-tagged recombinant proteins, purified and assembled into intact rod domains as described previously (Li et al., 2003). In pull-down experiments with rod domains bound to an affinity matrix and incubation with lysates from 3T3-DDR1 cells, DDR1 selectively bound to NMHC-IIA assembled into rods, but not to unassembled myosin or to NMHC-IIB (Fig. 2F). In separate pull-down experiments using lysates prepared from transfected HEK293 cells (which strongly expressed NMHC-IIA and -IIB when transfected) and performed in the same manner, NMHC-IIA rods bound to full-length DDR1b, independent of pre-treatment of the HEK293 cells with collagen (Fig. 2G). There was no binding to the truncated d-isoform or to DDR2 (Fig. 2G). These data suggest that DDR1 probably associates with the rod domain of NMHC-IIA in living cells and that intact DDR1 is required for this association.

Consistent with the immunoprecipitation studies of 3T3-DDR1 cells indicating an association between DDR1 and NMHC-IIA, confocal images of immunostained 3T3-DDR1 cells that had been allowed to spread for 1 hour on collagen showed colocalization of NMHC-IIA with DDR1 that was restricted to a central ring pattern (Fig. 3A). Actin filaments colocalized with NMHC-IIA along lamellipodial margins (Fig. 3B). In 3T3-DDR1 cells that migrated into an in vitro wound, DDR1 and NMHC-IIA colocalized at the leading cell edge (Fig. 3C). Quantitative assessment of the extent of colocalization in DDR1-expressing cells (by calculation of the correlation coefficient) showed significantly higher correlations of DDR1 and NMHC-IIA immunostaining for cells migrating on collagen than for cells migrating on tissue culture plastic or fibronectin (Fig. 3C, histogram). In spreading 3T3-DDR1 cells that had been immunostained, there was marked colocalization of DDR1 and β1 integrin at the cell peripheries, although the staining for DDR1 was more punctate than that of the β1 integrin (Fig. 3D).

We investigated the subcellular localization of NMHC-IIA in the context of DDR1 function by labeling 3T3-DDR1 and control cells with a NMHC-IIA-specific antibody (Fig. 4A). As previously reported, NMHC-IIA formed largely parallel arrays of filaments throughout the cell body but these were not present within lamellipodia (Even-Ram et al., 2007; Takubo et al., 2003). To assess the influence of collagen on the NMHC-IIA–DDR1 complex, we quantified changes in myosin filament bundling in 3T3-DDR1 and NIH3T3 cells (which do not express DDR1) after collagen stimulation. For NIH3T3 cells, plating on collagen caused no change in the percentage of myosin-filament-forming cells whereas NIH3T3 cells, expressing high levels of DDR1, exhibited much higher percentage of myosin-filament-forming cells (Fig. 4A), independent of substrate coating. There was a 20% increase in the number of 3T3-DDR1 forming myosin filaments when they were plated on collagen compared with those plated on tissue culture plastic (P<0.001). In mouse embryonic fibroblasts (MEF) derived from DDR1-knockout and wild-type animals, there was no difference in the percentage of myosin-filament-forming cells when plated on tissue culture plastic (Fig. 4B) but there was a significant increase in myosin-filament-forming wild-type cells when plated on collagen (20%; P<0.001).

We wondered whether plating cells on collagen would affect myosin function in relation to filament formation, specifically phosphorylation of serine 19 in the myosin light chain (Bresnick, 1999). 3T3-DDR1 cells were plated on collagen or tissue culture plastic and phosphoserine 19 of myosin light chain and β-actin (as loading control) were evaluated by immunoblotting (Fig. 4C). In cells plated on tissue culture plastic, there was moderate increase in phosphorylation at 10 minutes but there was more abundant phosphorylation, when adjusted for loading, at 30 and 60 minutes in cells plated on collagen. Phosphorylation of myosin light chain was much less in NIH3T3 cells, which do not express DDR1 (Fig. 4C, lower panels).

NMHC-IIA generates mechanical forces essential for cell migration (Even-Ram et al., 2007; Zeng et al., 2004). To determine whether DDR1 contributed to this process confluent cell monolayers were wounded and cell migration was measured 24 hours later. Migration of 3T3-DDR1 cells was more rapid than that of control NIH3T3 cells (Fig. 5A). The distance of the leading edge of migrating cells into in vitro wounds was quantified in cultures of 3T3-DDR1 and NIH3T3 cells, as well as MCF-7 cells treated with control or DDR1 siRNA (Fig. 5B). These data showed that in cells plated on collagen (but not on tissue culture plastic or fibronectin), expression of DDR1 was associated with significantly (P<0.05) greater migration distances. This migratory-enhancing function of DDR1 was confirmed with a Boyden-chamber cell migration assay employing uncoated or collagen-coated membranes and NIH3T3 and 3T3-DDR1 cells (Fig. 5C). Both NIH3T3 and 3T3-DDR1 cells migrated much more quickly on collagen than on uncoated membranes but the 3T3-DDR1 cells migrated more quickly than the NIH3T3 cells on collagen (P<0.05). Furthermore, there was increased migration of DDR1-expressing cells in an in vitro wounding assay using monolayer cultures of wild-type and DDR1-knockout mouse embryonic fibroblasts (Fig. 5D,E).

Blebbistatin, a specific inhibitor of the non-muscle myosin II ATPase (Straight et al., 2003), retards directional locomotion of a variety of cell types (Even-Ram et al., 2007). In migration assays of NIH3T3 and MCF-7 cells, as well as MEFs, we found that inhibition of NMHC-II ATPase activity with blebbistatin (25 μM) retarded migration only in those cells that expressed DDR1 (Fig. 6A; P<0.01). We visualized the assembly of myosin filaments by transfecting cells with a plasmid coding for a GFP-tagged NMHC-IIA, as previously described (Nizak et al., 2003). Fibroblasts from wild-type and DDR1-null mice, NIH3T3 and 3T3-DDR1 cells, and MCF-7 cells treated with control or DDR1 siRNA, were examined after plating on collagen and then the cell layers were wounded. GFP-NMHC-IIA-labeled filaments were readily imaged (Fig. 6B,C) and quantitative studies were conducted to estimate the velocity of labeled filament elongation as a measure of the rate of assembly of myosin filaments. Measurements were made in the leading edge of the cell and in the cell body. In cells expressing DDR1, the estimated filament elongation velocities were higher (P<0.05) near the leading edge of the cell (facing the in vitro wound) whereas in the cell bodies the velocities were higher in those cells expressing low or no DDR1 (P<0.01; Fig. 6D). These findings suggest a role for DDR1 in regulating the assembly of myosin IIA filaments.

We examined the role of actin filaments in the NMHC-IIA–DDR1 interaction by treating cells with cytochalasin D, a toxin that caps actin barbed ends and prevents actin filament assembly. In coimmunoprecipitation experiments, cytochalasin D enhanced interactions between NMHC-IIA and DDR1 (Fig. 6E) whereas cell lysates immunoprecipitated with pre-immune antibody showed no reactive NMHC-IIA or DDR1 (Fig. 6E). We examined the binding of collagen to DDR1 by incubating collagen-coated beads (with or without cytochalasin D or blebbistatin) with NIH3T3 and 3T3-DDR1 cells for 1 hour followed by immunoblotting of the collagen-bead-associated proteins. These data showed that blebbistatin and cytochalasin D treatments did not affect the relative abundance of DDR1 that was bound to the collagen beads (Fig. 6F). These findings are consistent with the notion that only a relatively small proportion of leading edge-localized NMHC-IIA associates with DDR1 during migration whereas after actin filament disassembly and disruption of actin-myosin interactions, additional NMHC-IIA can associate with DDR1. Collectively these data support the hypothesis that functional interactions between DDR1 and NMHC-IIA filaments are involved in cell migration.

We investigated the roles of DDR1 and NMHC-IIA in formation of initial cell-substratum contacts. We first determined whether NIH3T3 and 3T3-DDR1 cells would attach to collagen. Initial cell adhesion (10 minutes) to collagen was not affected by DDR1 expression (Fig. 7A). Cell spreading assays in time course experiments (10 minutes to 4 hours) were performed on collagen-coated plates with 3T3-DDR1 cells and mock-transfected NIH3T3 control cells, which do not have detectable levels of DDR1 (Fig. 1A,B). Compared with control NIH3T3 cells there was reduced (P<0.001) spreading of 3T3-DDR1 cells up to 2 hour; this difference in spreading dissipated by 4 hours (Fig. 7B,C). Similarly, in spreading assays using MEFs and MCF-7 cells treated with siRNA to DDR1, there was significantly (P<0.001) less spreading by those cells expressing DDR1 (Fig. 7B).

In experiments in which β1 integrin function was inhibited with an integrin-blocking antibody, NIH3T3 and 3T3-DDR1 cells were allowed to spread on collagen in the presence or absence of antibody for 1 or 4 hours (Fig. 7D). The integrin-blocking antibody inhibited cell spreading of NIH3T3 cells at 1 and 4 hours (P<0.001) whereas inhibition of spreading was only detectable in 3T3-DDR1 cells at 4 hours. In 3T3-DDR1 cells plated on collagen for 1 hour, inhibition of the NMHC-II ATPase by blebbistatin enhanced spreading (P<0.05) whereas there was no significant effect (P>0.2) on NIH3T3 cells (Fig. 7E). If blebbistatin treatment was followed by inhibition of β1 integrin function, with an integrin blocking antibody (1 hour), there was no further reduction of spreading of 3T3-DDR1 cells (Fig. 6E) whereas there was a marked inhibition of spreading of NIH3T3 cells.

Conceivably, the presence or absence of DDR1 may affect β1 integrin function. Accordingly, in non-permeabilized cells plated for 1 hour on collagen, fibronectin or tissue culture plastic, we measured β1 integrin activation with 9EG7, an activation-specific, neo-epitope antibody (Lenter et al., 1993) that binds to activated β1 integrins. By quantitative fluorescence photometry of individual cells we found no effect of DDR1 expression on β1 integrin activation on cells plated on collagen, fibronectin or tissue culture plastic (P>0.2; Fig. 7F), which is consistent with earlier data (Vogel et al., 2000) but is at variance with a more recent report (Yeh et al., 2009).

The discoidin domain receptors, DDR1 and DDR2, are a sub-family of receptor tyrosine kinases that function as collagen receptors and may act independently of the β1 integrin (Vogel et al., 2000). Both DDRs are important regulators of a wide variety of developmental processes and can modify cell adhesion, the migration of cells over various substrates, proliferation and the remodeling of the extracellular matrix (Vogel et al., 2006; Leitinger and Hohenester, 2007). The proteins that interact with DDRs to mediate these processes are poorly defined. The main findings of this study are that DDR1 and NMHC-IIA associate with one another, and that DDR1 and non-muscle myosin activity regulate cell spreading and migration on collagen.

The association between DDR1 and NMHC-IIA was demonstrated by coimmunoprecipitation of cell lysates and by in vitro binding assays using recombinant rod domains of NMHC-IIA with DDR1 prepared from cell lysates. This association increased after cell binding to collagen and occurred prior to collagen-induced phosphorylation of DDR1 (Vogel et al., 1997). Consistent with these data, two-color confocal microscopy and correlation analyses of intact cells showed that DDR1 and non-muscle myosin IIA colocalized. The regulation of the DDR1–NMHC-IIA association apparently relies on intact NMHC-IIA, since we did not detect collagen-induced changes in the affinity of activated DDR1 for the isolated rod domain of NMHC-IIA. Notably, a direct interaction has been demonstrated between the NMHC-IIA rod domain and S100A4 (Li et al., 2003) and with the small GTPase Rap1 in regulation of collagen binding to the β1 integrin (Arora et al., 2008), but it is not known how the association of NMHC-IIA with DDR1 is regulated by collagen adhesion.

Our finding that depolymerization of actin filaments with cytochalasin D strongly enhanced the association of DDR1 with NMHC-IIA suggests that this association may be regulated in part by the interaction of NMHC-IIA with intact actin filaments. Previous studies have shown that the non-muscle myosin interactions with actin that are required for force-generating structures are dissipated by cytochalasin (Kolega and Kumar, 1999). Conceivably, in early stages of cell adhesion to collagen, DDR1 does not strongly associate with NMHC-IIA because much of the myosin is already actin-associated and is not accessible for interaction with DDR1. As shown by our data, within 6 hours of cell adhesion to collagen, non-muscle myosin IIA associates with DDR1. Possibly, this association is mediated by remodeling of actin filaments, a process that is induced by cell adhesion (Mooney et al., 1995) and that may enable dissociation of actin from myosin and thereby enhance DDR1-non-muscle myosin interactions.

Our data showed that NMHC-IIA interacts with full-length DDR1, but not with the DDR1d isoform that lacks most of the cytoplasmic region. Five isoforms of DDR1 are generated by alternative splicing, which lead to deletions of the juxta-membrane or kinase domains (Alves et al., 2001). Conceivably, the association of NMHC-IIA with DDR1 and the possibly resultant effect on cell spreading and migration depend on full-length DDR1. In this context, full-length a- or b-isoforms of DDR1 enhance leukocyte adherence to collagen (Kamohara et al., 2001) and our data show that full-length DDR1, when transfected into cells, increases migration over collagen.

Cognizant of the complexity of the processes regulating adherence and migration, we examined the effect of DDR1 alone on cell attachment, spreading and migration. Although DDR1 expression had no effect on initial cell attachment, it strongly inhibited cell spreading. Thus whereas NIH3T3 control cells rapidly extended cell protrusions on collagen-coated dishes, 3T3-DDR1 cells took much longer to become fully spread. A recent report has indicated that DDR1 inhibits cell spreading by suppressing α2β1-integrin-mediated activation of Cdc42 (Yeh et al., 2009) and that a signaling complex comprising DDR1 and SHP-2 inhibits α2β1-integrin-mediated cell migration (Wang et al., 2006). Our data indicate that whereas DDR1 and β1 integrins co-distribute at the leading edges of cells, the expression of DDR1 restricts cell spreading, an effect that is relieved by the non-muscle myosin II inhibitor, blebbistatin, consistent with earlier findings (Even-Ram et al., 2007). Furthermore, our data are consistent with the notion that following initial adherence of cells to collagen surfaces via DDR1, myosin IIA-dependent contractility prevents cell spreading (Sandquist et al., 2006). In NIH3T3 cells in which DDR1 was not expressed and adherence to collagen relied on β1 integrins and presumably other collagen receptors, blebbistatin had no effect on cell spreading. Therefore the effect of DDR1 expression in restricting cell spreading on collagen was dependent on non-muscle myosin II-dependent contractility. This contention is consistent with our observation of prolonged myosin light chain phosphorylation at serine 19 in 3T3-DDR1 cells spreading on collagen in comparison to cells spreading on tissue culture plastic. Indeed, assembly of myosin II into cytoskeletal structures, where it can generate and resist forces, is regulated by phosphorylation of myosin light chains at serine 19 (Kolega and Kumar, 1999). These findings do not rule out an independent process by which DDR1 may inhibit Cdc42-induced filopodia formation (Yeh et al., 2009), an important initial step for cell spreading (Partridge and Marcantonio, 2006).

In contrast to inhibition of non-directional cell spreading, DDR1 enhanced directed cell migration on collagen, as we observed in Boyden chamber assays and in in vitro wounds. In all of the cell types that were examined, expression of DDR1 enhanced directed cell migration for cells plated on collagen but much less so for cells plated on fibronectin or tissue culture plastic. This effect may depend, in part, on non-muscle myosin II since treatment of DDR1-expressing cells (but not DDR1 null cells) with blebbistatin reduced cell migration. Furthermore, the rate of assembly of non-muscle myosin filaments was much more rapid at the leading edge of cells expressing DDR1 than cells that were null for DDR1. Taken together these data indicate that once DDR1-expressing cells make sufficient numbers of contacts with the substratum and spread fully, DDR1 and NMHC-IIA facilitate the process of directional cell migration. Previous data have shown that blebbistatin enhances cell migration (Even-Ram et al., 2007) but the expression of DDR1 in these cells was not described.

We found that DDR1-expressing cells plated on collagen generated more prominent myosin filaments than DDR1 null cells, therefore DDR1 may enhance the generation of increased contractile forces while simultaneously dampening the protrusive forces associated with cell spreading. Conceivably, in the absence of DDR1-collagen attachments, in fibroblasts plated on tissue culture plastic or in cells null for DDR1, there is more limited adhesion strength and consequently protrusive forces may exceed contractile forces. Under these conditions relatively low myosin activity would be associated with the generation of weak cytoskeletal tensile forces (isometric contractility) and cell migration would be limited. By contrast, fibroblasts plated on collagen use DDR1 to more effectively `pull' on the substratum, and stronger myosin-driven contractile forces overshadow protrusive forces, enabling faster migration rates. This notion is consistent with our observation that non-muscle myosin filament assembly was much more rapid in the leading edge of cells expressing DDR1 than in DDR1 null cells.

In summary, our data provide evidence for a novel association between DDR1 and NMHC-IIA. This association may be important for regulating the generation of contractile forces that are important for cell migration and spreading. The association between DDR1 and NMHC-IIA could constitute part of a larger signaling network analogous to the β1 integrin-focal adhesion kinase signaling pathway and may be of fundamental importance for wound healing and organ regeneration.

Mouse NIH3T3 fibroblasts (from ATCC) and primary mouse embryonic fibroblasts (MEF, from day 12.5 embryos) were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum and penicillin and streptomycin (100 U/ml and 100 μg/ml). The expression of DDR2 was not altered in DDR1-null cells compared with control cells. These cells provided a good in vitro system to compare the effect of the presence or absence of DDR1 on cell migration and spreading, and are readily transfected. NIH3T3 cells were stably transfected with the DDR1 (b-isoform) or a control (pLXSN) plasmid (Vogel et al., 2000). Human breast cancer MCF-7 and MDA-MB-231 cells (ATCC) were grown in RPMI or DMEM medium with 10% fetal calf serum, respectively. These cells bind to collagen tightly, can invade collagen and are readily transfected, thereby facilitating protocols that required the modulation of DDR1 levels. Integrin-β1-deficient GD25 cells were provided by Reinhard Fässler (Max-Planck Institute for Biochemistry, Munich, Germany). As these cells do not express the β1 integrin and are readily transfected with DDR1-expressing plasmids, they facilitate experiments examining the functional importance of DDR1 and the β1 integrin. Human embryonic kidney 293 (HEK293) cells were used for expression studies of DDR1 and non-muscle myosin IIA interaction experiments. Plasmids coding for NMHC-IIA and NMHC-IIB rod domains were kindly provided by Anne Bresnick (Albert Einstein College of Medicine, New York). DDR1-specific and control siRNA pools were purchased from Dharmacon and transfected according to the manufacturer's instructions. Cytochalasin D and (±)-blebbistatin were obtained from Calbiochem. A monoclonal antibody directed against DDR1 (YH1-10D4) was generated by injecting DDR1-expressing C2C12 cells into BALB/c mice, and was used for immunofluorescence staining. DDR1 was identified by immunofluorescence using a biotinylated anti-DDR1 (BAF2396, R&D Systems, Minneapolis, MN). A polyclonal antibody raised against a peptide corresponding to amino acids 505-523 of DDR1b was described previously (Vogel, 2002). The following antibodies were employed for immunoblotting and/or immunostaining: anti-DDR1 and anti-β1-integrin (sc-532 and N-20 respectively, Santa Cruz Biotechnology; 9EG7-BD Biosciences), anti-β-actin (clone AC15; Sigma-Aldrich), anti-NMHC-IIA (Biomedical Technologies) and anti-phosphoserine 19 of myosin light chain (Cell Signaling).

Cells were lysed on ice in lysis buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1.5 mM MgCl₂, 5 mM EGTA, 10% glycerol, 1% Triton X-100, 10 μg/ml aprotinin, 10 mM sodium fluoride, 1 mM PMSF and 1 mM sodium orthovanadate) and lysates were centrifuged at 14,000 g at 4°C for 10 minutes. For immunoprecipitation, 1 mg of solubilized protein was incubated with 0.5 μg polyclonal anti-DDR1 (Vogel, 2002), 0.1 μg of biotinylated polyclonal anti-DDR1 antibody (R&D systems) or anti-NMHC-IIA antibody (Biomedical Technologies), along with protein A-Sepharose beads (Amersham Biosciences) at 4°C overnight. Beads were washed three times with buffer containing 20 mM Hepes pH 7.5, 150 mM NaCl, 0.1% Triton X-100, and 10% glycerol. Proteins were eluted with Laemmli buffer at 95°C for 5 minutes and separated by SDS-PAGE. Western blots were developed using enhanced chemoluminescence (Amersham). All western blot experiments were repeated at least three times and representative data sets are presented. His-tagged proteins coding for residues 1339-1961 of the human myosin-IIA heavy chain or residues 1346-1976 of the human myosin-IIB heavy chain were isolated from bacterial cultures and assembled into functional myosin rod domains as described previously (Li et al., 2003). For analysis of collagen-bead-associated proteins, cells were incubated with collagen beads and 1 hour later, cells were lysed and collagen beads were separated. Bead-associated proteins were eluted and immunoblotted (Arora et al., 2008).

Immunoprecipitated protein complexes were resolved by SDS-PAGE and visualized with Coomassie Blue. Protein bands were excised and digested with trypsin (Roche Diagnostics). Saturated α-cyano-4-hydroxycinnamic acid in 70% acetonitrile/0.1% trifluoroacetic acid was used as the matrix solution. The peptide mix was subjected to matrix-assisted laser desorption ionization time-of-flight analysis on a Voyager-DE STR mass spectrometer (Applied Biosystems). The mass spectra were externally calibrated from the molecular mass of a mixture of standard peptides. Results were analyzed using ProFound (http://129.85.19.192/profound_bin/WebProFound.exe).

Cells grown on glass coverslips were washed three times with PBS, fixed in 2.5% paraformaldehyde for 10 minutes, washed again with PBS, permeabilized with 0.1% Triton X-100 for 5 minutes and blocked with 1% BSA in PBS at room temperature for 1 hour. Cells were incubated with primary antibodies at room temperature for 1 hour, extensively washed, and incubated with secondary FITC-labeled anti-rabbit antibody (Sigma-Aldrich), Cy3-labeled anti-mouse antibody (Jackson Immuno Research Laboratories, Inc.) or Cy3-labeled streptavidin (Sigma) at room temperature for 1 hour. Coverslips were mounted with 33% glycerol and 13% Mowiol 4-88 (Calbiochem). Images were captured using a LSM510 Zeiss laser scanning confocal microscope or a Leica TCS confocal microscope. Colocalization of immunofluorescence images was analyzed in six fields per cell using the ImageJ plug-in JACoP (Bolte and Cordelieres, 2006). The Pearson's correlation coefficient was expressed as the mean ± s.d.

Glass coverslips were coated with 10 μg/ml type I collagen (from rat tail; BD Biosciences) that was allowed to polymerize, incubated at 4°C overnight and washed with PBS twice prior to use. The efficacy of the collagen coating was verified by immunostaining for collagen and by scratch wounding to visualize the denuded surface. Cells were harvested by briefly exposing the cells to 0.05% trypsin in EDTA, following extensive washing with medium plus 10% serum. In contrast to stimulation with collagen, serum-containing medium had no effect on DDR1 activation (Vogel et al., 1997). Cells were loaded onto coverslips in medium with 1% serum for the indicated period of time. Non-adherent cells were removed, and adherent cells were stained with either Diff-Quik (Invitrogen, Canada) or 5 μg/ml Alexa-Fluor-488-conjugated wheat germ agglutinin (WGA; Molecular Probes). Following Diff-Quik staining, five randomly chosen fields were counted, and spreading cells were defined as those cells with total cell size twice the size of the nucleus. Following WGA staining, cell images in five random fields were captured by confocal microscopy and cell surface area was measured using ImageJ software (version 1.37, NIH). To measure migration, 1×10⁴ cells were seeded on top of a Transwell insert (8 μm pore size; Costar) coated with type I collagen or left uncoated, and 50 μg/ml of collagen was added as a chemoattractant to the lower reservoir. After 18 hours, cells migrating to the underside of the filter were counted following Diff-Quik staining. All experiments were performed in triplicate and repeated at least three times.

Cells (5×10⁴) were loaded onto 96-well plates in medium with 1% serum for the indicated period of time; unattached cells were removed by washing with PBS. Cells were cultured with 0.05% MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] in medium with 1% serum for 3 hours. After removing the medium, 200 μl DMSO was added to each well and absorbance measured at 570 nm. All experiments were performed in triplicate.

Cells were cultured on tissue culture plates until confluent. An in vitro `wound', generated by scratching with the blunt end of a 200 μl plastic pipette tip, was created and the edge of the wound was marked. After washing with PBS, cells were cultured in the medium with 10% serum in the absence or the presence of blebbistatin (50 μM) or DMSO for 18 hours. The average distance of cell migration from the initial wound edge was measured at ten random loci using Zeiss Axiovision software. Similarly, equal numbers of DDR1-null and control MEFs were plated on collagen-coated dishes and wounded by gently scratching with a pipette tip. Wound closure was followed by videomicroscopy, taking one frame every 6 minutes for 6 hours (Nikon 200 microscope and SimplePSI software). For imaging of living cells, MEFs were transfected with the plasmid pEGFP-SF9, which encodes an EGFP-tagged NMHC-IIA (Nizak et al., 2003). Confluent cells were wounded and the organization of endogenous myosin filaments were recorded using a Zeiss 200M microscope equipped with a Solarmere Yokogawa spinning disk confocal scanning system and a 63× Plan-Apochromat objective, which was placed in an environmental chamber maintained at 37°C, 5% CO₂ and at constant humidity. Images were captured every 3 seconds for a total duration of 8 minutes using a XR/Mega10 CCD camera (Stanford Photonics). Deconvolution of the images was performed using AutoDeblur (Media Cybernetics). The rate of assembly of individual GFP-labeled non-muscle myosin filaments (velocity expressed in μm/seconds) was quantified in the leading edge of cells and in cell bodies using ImagePro Plus Version 6.1 (Media Cybernetics). Transfection with a pEGFP control plasmid yielded no specific cellular staining.

All data are presented as mean value ± standard deviation of the mean (s.d.). Student's t-test was used for statistical comparison of mean values between two groups. Differences between means with P<0.05 were considered to be statistically significant.