Differential Function of N-Cadherin and Cadherin-7 in the Control of Embryonic Cell Motility

FN, laminin I, vitronectin, type I and type IV collagens, ECM gel, FITC-phalloidin, RGDS peptide, control rat IgG, the mouse mAb (clone GC-4) directed against the A-cell adhesion molecule (A-CAM, also known as N-cadherin), and the rabbit polyclonal anti–pan-cadherin antibody were purchased from Sigma Chemical Co. Mouse mAbs directed against β-catenin (clone 14) and against FAK (clone 77) were obtained from Transduction Laboratories. The mouse mAb directed against phosphotyrosine (clone 4G10) was obtained from Upstate Biotechnologies. The rat mAb directed against mouse β1 integrin (clone 9EG7) was purchased from PharMingen. The NC-1 mouse mAb against avian NCC has been previously described elsewhere (Vincent et al., 1983). The mouse mAb (clone F-VII) against human vinculin was generously donated by M. Glukhova (Institut Curie, France). The rat mAb NCD-2 and the mouse mAb CCD7-1 were kindly donated by M. Takeichi and S. Nakagawa (Kyoto University, Kyoto, Japan) and are directed against chicken N-cadherin and cadherin-7, respectively. Secondary antibodies coupled to Texas red and horseradish peroxidase were purchased from Amersham.

The S180 cells used in this study were a subclone derived by Dr. K. Yamada (National Institutes of Health, Bethesda, MD) from the original parental cell line, which was obtained from American Type Culture Collection. These cells were selected for their inability to assemble a FN matrix at the cell surface. They were cultured in DMEM containing 10% (vol/vol) FCS, penicillin (100 IU/ml)/streptomycin (100 μg/ml), and 2 mM l-glutamine (Seromed) in a 37°C incubator under an atmosphere of 6% CO₂/94% air. The Ncad-1 expressor clone for chicken N-cadherin and the cad7-29 expressor clone for chicken cadherin-7 were kindly produced by S. Nakagawa. In brief, S180 cells were transfected with either pMiwcCad7 (Nakagawa and Takeichi, 1995) or pMiwcN (Fujimori and Takeichi, 1993) together with pGKNeoB by the calciumphosphate precipitation method. Transfectants were selected in culture medium containing G418 (Life Technologies).

Assays of cell adhesion to FN-coated substrates were performed on bacteriological petri dishes. Droplets of FN (100 μl of 0.01–50 μg/ml in PBS containing 10 μg/ml BSA) were deposited on bacteriological dishes and incubated overnight at 4°C followed by a 30-min incubation with 3 mg/ml BSA in PBS (previously heat-inactivated for 3 min at 80°C). The substrates were thoroughly washed and maintained in PBS until use. S180, cad7-29, and Ncad-1 cells were harvested with cell dissociation enzyme-free buffer (Life Technologies) for 10 min at 37°C. Cells were pelleted by centrifugation and incubated for 45 min in DMEM containing with 10% (vol/vol) FN-depleted FCS + 1 mM EGTA to prevent self cell-cell aggregation. Cells were then centrifuged, resuspended at a density of 2.5 × 10⁵ cells/ml, and 100-μl droplets of the suspension were deposited on the pre-coated substrates. The dishes were incubated at 37°C for 1 h, rinsed with PBS to remove the nonadherent cells, and fixed in 5% glutaraldehyde in PBS. Cells were observed and images were recorded using a Nikon phase-contrast microscope equipped with a high performance CCD camera. The images were transferred to a Power Macintosh G3 computer. Cell adhesion was then quantified by colorimetric analysis. Cells were stained with 1% crystal violet in 200 mM MES and destained in 200 mM MES buffer, pH 6. The crystal violet fixed to the cells was dissolved in 10% acetic acid and OD was measured at 570 mM. The control values correspond to the OD obtained for cells deposited on BSA. 100% cell adhesion corresponds to the value obtained on FN-coated substrate at 50 μg/ml. The extent of cell spreading was quantified using Scion Image software by measuring the area of cells interacting with the substrate (at least 30 cells were measured). At least three independent experiments were done for each FN concentration and for each cell type.

S180, cad7-29, and Ncad-1 cells were plated in culture 1 d before the assay. They were harvested from monolayer cultures as previously described (Nakagawa and Takeichi, 1995) to preserve cadherins at the cell surface. 2 × 10⁶ cells were added to an Erlenmeyer flask containing 5 ml of HMF (10 mM Hepes, magnesium-free PBS, pH 7.4, 10 mM CaCl₂). The cell suspensions were incubated at 37°C in a gyratory shaker at 75 rpm. The kinetics of cell aggregation were determined by collecting several aliquots of the cell suspension at various times. The degree of cell aggregation was estimated according to the decrease in particle number: percent aggregation = 1 − [number of particles Px (at time t = x)/initial particle number P0] × 100. At least three independent experiments were done for S180, cad7-29, and Ncad-1 cells.

Before grafting experiments or aggregation assay, cells were labeled with fluorogold as previously described (Beauvais et al., 1995). In brief, 50% confluent cell cultures were incubated for 4 h in 4 ml of culture medium containing 24 μl of fluorogold stock solution (2% in Locke’s saline; Fluo-rochrome Inc.). The cells were washed three times in Hanks’ balanced salt solution (Life Technologies) to remove all residual label. Cells were harvested in a solution of trypsin/EDTA, centrifuged, and resuspended in fresh culture medium.

5 × 10⁶ cells were added to an Erlenmeyer flask containing 5 ml of DMEM supplemented with 10% (vol/vol) FCS, previously equilibrated in a 6% CO₂/94% air atmosphere. The cell suspensions were incubated at 37°C in a gyratory shaker at 75 rpm for 24 h.

Cell aggregates were dispersed on bacteriological petri dishes or glass coverslips coated with 100-μl droplets of FN (10 μg/ml), laminin (50 μg/ml), type I collagen (30 μg/ml), type IV collagen (30 μg/ml), or vitronectin (10 μg/ml) in PBS and were incubated at 37°C for 1 h. Coated substrates were rinsed twice in PBS, saturated with BSA, and extensively washed. Aggregates were then deposited and allowed to disperse before fixation.

Aggregates were cultured in a three-dimensional environment, by allowing a mixture of one part ECM gel to one part DMEM (vol/vol) containing a few aggregates to gel at 37°C for 30 min before incubation with the culture medium.

For the mixed cell aggregation assay, Ncad-1 cells and fluorogold-labeled cad7-29 cells were mixed in an Erlenmeyer flask in a 1:1 ratio at a total cell density of 5 × 10 ⁶ cells/ml. They were incubated in a gyratory shaker at 37°C for 16 h. Aggregates were fixed by incubation with 0.5% glutaraldehyde for 5 min, plated on glass coverslips, and observed by fluorescence microscopy.

The dispersion of aggregates was observed at various times with a Ni-kon inverted phase-contrast microscope. In some experiments, dispersing aggregates were incubated for 1 h with 100 μg/ml of inhibitory rat monoclonal anti–mouse β1 integrin, control rat IgG, or 2 mg/ml RGDS peptides with or without cycloheximide (CHX, 10 μg/ml) before fixation.

Cell aggregates were deposited on an FN-coated bacteriological petri dish and incubated at 37°C in culture medium. The plate was then placed on the stage of a computer-controlled epifluorescence microscope (Leica DMIRBE) equipped with an enclosed warming incubator, a cooled CCD camera (Princeton RTE/CCD), and a PC workstation. The enclosed incubator made it possible to maintain the sample at 37°C in a humidified atmosphere containing 6% CO₂/94% air. Metamorph software (Universal Imaging) controlled image acquisition, light intensity, fluorescence shutter, filter wheel, and motorized stage. Images were recorded every 4 min. Migratory cell paths were traced and the speed of locomotion was calculated by dividing the total distance of migration by the time of the experiment. This was done for at least 20 cells in two independent experiments.

Cell migration was quantified by determining the extent of migration from agarose droplets using a modification of the method described by Varani et al. (1978). In brief, cells were harvested by trypsin-EDTA treatment and centrifugation and were resuspended at a density of 3.3 × 10⁷ cells/ml in CO₂-independent medium (Life Technologies) containing 0.2% low-melting agarose (Sea Plaque, FMC Bioproducts, TEBU) maintained at 39°C. The cells were plated as 0.8-μl droplets on flat-bottomed wells (microtest III flexible 96-well assay plates, non-tissue culture–treated; Falcon) precoated with 10 μg/ml FN and the agarose was allowed to solidify for a few minutes at 4°C. Culture medium was added with care to prevent the detachment of droplets. Cells were cultured at 37°C for 20 h and fixed with 3.7% formaldehyde. Cells were rinsed, stained with crystal violet, and rinsed again. Images were recorded using a Nikon phase-contrast microscope equipped with a high performance CCD camera. They were transferred to a Power Macintosh G3 computer. The total area of cell outgrowth was measured directly on the monitor for at least 24 droplets for each cell type. The results are expressed in arbitrary units.

Cells were rinsed twice in PBS containing 2 mM CaCl₂ and MgCl₂ and incubated in methanol for 10 min at −20°C. They were then incubated with acetone for 2 min and air dried. The cells were rehydrated in PBS and treated with blocking solution (3% BSA in PBS) for 30 min. They were then stained for 1 h at room temperature with primary antibody diluted in blocking solution. Cells were rinsed twice in PBS and incubated with secondary antibody coupled to Texas red for 45 min. Before immunofluorescent labeling with FITC-phalloidin or antivinculin antibody, the cells were fixed for 20 min in 4% paraformaldehyde in PBS, permeabilized by incubation for 5 min in 0.5% Triton X-100 in PBS, rinsed three times, incubated for 30 min in 20 mM glycine in PBS, and treated with blocking solution (3% BSA in PBS) for 30 min to reduce background signal. Preparations were observed using a motorized epifluorescence microscope (Leica DMRBE) equipped with a cooled CCD camera (Hamamatsu C5985). Acquisitions were controlled by a Power Macintosh workstation via IP-Lab software. Images were recorded at the same time as exposure and gain values to make it possible to compare in fluorescence intensity.

For Western blot analysis of β-catenin or cadherins, subconfluent monolayers of cells were grown in standard conditions or in culture medium containing 10 μg/ml of CHX for 2–5 h before extraction. Cell extracts were prepared in situ by incubation with 1 ml/petri dish of lysis buffer (1% Triton X-100, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 2 mM PMSF, 2 μg/ml leupeptin, 2 μg/ml aprotinin in PBS containing 2 mM CaCl₂ and MgCl₂) for 15 min on ice.

For Western blot analysis of tyrosine-phosphorylated proteins, cell extracts were prepared in situ by incubation with lysis buffer (1% NP-40, 150 mM NaCl, 50 mM Tris, pH 8, 5 mM EDTA, 2 mM PMSF, 2 μg/ml leupeptin, 2 μg/ml aprotinin, 100 μM sodium orthovanadate) for 15 min on ice. Four different treatments were applied to cells before extraction. Cells were harvested as described for the cell-cell adhesion assay and were allowed to aggregate in HMF for 2 h to produce well defined aggregates (except for S180 cells which cannot aggregate). They were centrifuged before extraction (lane 1 of Western blot). Alternatively, cells were incubated in suspension for 10 min (Ncad-1) or 30 min (cad7-29 and S180) to ensure that most of the cells were engaged in the initial intercellular aggregation process (small aggregates, except for S180 cells which cannot aggregate). Cells were then plated for 30 min on FN- (10 μg/ml; lane 3) or BSA-coated substrate (lane 4) before extraction. Alternatively, cells were harvested as described for the cell-FN adhesion assay and plated for 30 min on FN-coated substrate (at a density preventing cell-cell adhesion) before cell extraction (lane 2).

Cells were incubated with lysis buffer on ice, scraped from the petri dish with a rubber policeman, homogenized, and centrifuged for 15 min at 20,000 g at 4°C. Supernatants were collected and protein content determined by protein assay (Bio-Rad). Proteins (50 μg in SDS sample buffer) were subjected to electrophoresis in 7.5% or 10% acrylamide gels. Proteins were transferred electrophoretically from gels to Immobilon-P filters (Millipore). The membranes were incubated with appropriate antibodies. In brief, membranes were incubated with blocking solution (0.1% gelatin, 0.1% Tween 20 in PBS) for 1 h on a gyratory shaker at room temperature and were rinsed four times in PBS containing 0.1% Tween 20. They were incubated overnight at 4°C with anti–β-catenin (1:2,000), CCD7.1 ascites (1:4,000), anti–pancadherin serum (1:2,000), anti-FAK (1:1,000), or antiphosphotyrosine (4G10; 1:1,000) in GT-PBS. Membranes were thoroughly washed and incubated for 1 h with anti–mouse or anti–rabbit IgG coupled to horseradish peroxidase at a dilution of 1:10,000. Membranes were washed and incubated with a chemiluminescence detection reagent (Amersham) for 1 min. The reagent was drained off and the membranes were placed against Hyperfilm (Amersham).

The Ncad-1 and cad7-29 cells were labeled with fluorogold, treated with trypsin-EDTA solution, centrifuged, and allowed to form small aggregates for 4 h. The aggregates were rinsed in sterile Locke’s saline medium (LSM). They were then transferred to a petri dish containing LSM, and were lightly stained with a few drops of 0.2% neutral red.

The parental S180 cells were labeled with fluorogold, treated with trypsin-EDTA solution, and centrifuged. The cell pellet was resuspended in fresh culture medium and incubated for 2 h in a 37°C water bath to allow the cells to reconstitute membrane proteins. The cells were repelleted by centrifugation. The pellet was transferred to a petri dish containing LSM, and was lightly stained with a few drops of 0.2% neutral red.

White Leghorn chicken embryos were incubated at 38°C until they had developed 20–23 somite pairs, corresponding to stage 13–14 (according to Hamburger and Hamilton, 1951). Grafting experiments were performed as previously described (Beauvais et al., 1995). A small aggregate or cell pellet was inserted into the embryo with a Spemann pipette. It was gently maneuvered with a tungsten needle through a slit in the ectoderm between the neural tube and somite in the area to which NCC migrate after leaving the dorsal surface of the neural tube (i.e., the seventh somites anterior to the last-formed somite; according to Loring and Erickson, 1987). The relocation of the graft site was facilitated by inserting a small amount of black charcoal into the center of the second somite posterior to the graft site. The embryo was then carefully picked up with a ring of filter paper and suspended on a nucleopore filter (polycarbonate, 8-μm pores) over the well of an organ culture dish (Falcon) filled with L15 medium (Life Technologies) containing 10% FCS. The embryos were returned to the 38°C incubator for a further 18 h and were then fixed. Grafts for 48-h grafting experiments were done in a similar manner, but directly in ovo after opening the shell with scissors and making a hole in the vitelline membrane as described by Selleck (1996).

Embryos were fixed in 4% paraformaldehyde for 2 h, washed with PBS, and a region of the embryo six somites in length, encompassing the graft, was cut out with tungsten needles. Embryo pieces were washed for 3 h in blocking solution and incubated overnight at 4°C in 0.5% BSA in PBS containing NC-1 antibody to label the NCC. Unbound antibody was removed by a 3-h incubation in 0.5% BSA in PBS and the specimens were immersed overnight in the secondary antibody. Specimens were rinsed several times and postfixed by incubation in 0.1% paraformaldehyde for 1 h and rinsed twice in PBS. The explants were then dehydrated, incubated twice for 3 min each in xylene, embedded in paraplast, and sectioned as previously described (Levi et al., 1987).

N-Cadherin and cadherin-7 belong to different cadherin subfamilies, type I and II, respectively. They are specifically regulated during NCC migration. N-Cadherin is expressed by NCC resident in the neuroepithelium and after aggregation to form ganglia, whereas cadherin-7 starts to be synthesized only when NCC begin to migrate. The spatio-temporal distribution of these cadherin subtypes during early embryogenesis suggests different functions in cell migration. We investigated the role of these cadherins in the control of in vitro and in vivo cell motility using S180 cells genetically modified by stable transfection so as to synthesize chicken N-cadherin or cadherin-7. S180 cells were used because they have a fibroblast-like morphology and no cell-cell adhesion properties. These cells are able to migrate after grafting into the embryonic environment (Erickson et al., 1980). They are a good model system for investigating the role of cadherins in the control of cell motility in the embryo.

Stable S180 transfectants expressing cadherin-7 (Fig. 1 B) or N-cadherin (Fig. 1 C) exhibited morphological differences from parental S180 cells (Fig. 1 A) characteristic of a cell-cell adhesive phenotype. The two types of transfectants were more flattened than parental cells. Ncad-1 cells gave rise to clusters of cells that were more cohesive that those of cad7-29 cells. Cadherin-7 (Fig. 1 D) and N-cadherin (Fig. 1 E) are present along cell boundaries, as shown by fluorescence immunostaining with specific antibodies CCD7-1 and NCD2, respectively.

Ncad-1 and cad7-29 cells, harvested from tissue culture dishes and put into suspension, form aggregates. Fluorogold-labeled cad7-29 cells were mixed with unlabeled Ncad-1 cells and allowed to aggregate for 16 h. Each clone segregated to form separate cohesive aggregates that could be distinguished under UV illumination (Fig. 1 F). Unlabeled aggregates were sometimes found apposed to labeled aggregates but no mixed aggregates were formed. This phenomenon is typical of a homotypic aggregation and shows that N-cadherin cannot interact with cadherin-7 as previously observed for L cells (Nakagawa and Takeichi, 1995). We analyzed cadherin expression in each transfectant by Western blot analysis of β-catenin content. We did this to determine whether β-catenin was similarly up-regulated relative to the endogenous level of β-catenin in parental cells (Fig. 1 G). S180 cells had very low levels of β-catenin which was only detectable on overexposed autoradiographs. The expression of either cadherin-7 or N-cadherin induced a strong increase in the amount of β-catenin. We performed several Western blot analyses and observed similar or slightly higher levels of β-catenin in Ncad-1 cells (see also Fig. 9). This suggests that cad7-29 and Ncad-1 cells have similar amounts of cadherin-7 and N-cadherin, respectively, at the cell surface.

Cell-cell adhesion assays were performed to compare the intercellular adhesion properties of parental and transfected cells (Fig. 2). During the initial phase of the aggregation process, a major difference was observed between Ncad-1 and cad7-29 cells. 70% of cell adhesion was achieved by Ncad-1 cells within the first 15 min, whereas the same extent of cell aggregation took 1 h to achieve for cad7-29 cells (Fig. 2 A). Ncad-1 and cad7-29 cells reached their maximum level of cell adhesion after ∼4 h, whereas parental cells did not aggregate (Fig. 2 D). At that time and earlier during the aggregation process, Ncad-1 cells formed larger aggregates (Fig. 2, B and C) than cad7-29 cells (Fig. 2, E and F). Cell aggregates continued to grow in size thereafter but after 24 h there is a little difference between Ncad-1 and cad7-29 aggregate sizes (see Fig. 3, A and F).

We investigated the response of transfected cells to the ECM environment by analyzing the ability of cad7-29 (Fig. 3, A–E, and Fig. 4 A) and Ncad-1 aggregates (Fig. 3, F–J, and Fig. 4 C) to spread and disperse in vitro on two-dimensional substrates coated with FN, laminin-1, type I collagen, type IV collagen, or vitronectin and within a three-dimensional ECM gel (composed mostly of laminin-1, type IV collagen, nidogen, and heparan sulfate proteoglycans).

cad7-29 aggregates were totally dispersed on FN after 5 h (Fig. 3 B). They were less efficiently dispersed on laminin-1 (Fig. 3 C) and on vitronectin (Fig. 3 D). Ncad-1 aggregates also dispersed on FN (Fig. 3 G) and on laminin-1 (Fig. 3 H) but to a lesser extent than observed for cad7-29 aggregates, whereas dispersion on vitronectin was inefficient (Fig. 3 I). After 5 h of culture, aggregates of both cad7-29 (Fig. 4 A) and Ncad-1 (Fig. 4 C) adhered but could not disperse on type I collagen. In contrast, the two types of aggregates did not adhere to type IV collagen within the first 5 h of culture (not shown). However, isolated S180, Ncad-1, and cad7-29 cells adhere to both type I and type IV collagen-coated substrates within an hour, but could not spread on them in standard cell-substrate adhesion assays (not shown). Thus, the greatest difference in scattering between Ncad-1 and cad7-29 aggregates was observed on FN. Ncad-1 cells migrating away from aggregates were tightly apposed to each other, whereas cad7-29 cells were more loosely connected and were sometimes found as individual cells.

We quantified the dispersion of Ncad-1 and cad7-29 aggregates by measuring the area of 10 aggregate outgrowths at various times (t = x) and calculating the mean area (O). The ratio O(t = x)/O(t = 0) was calculated for the two types of aggregate deposited on each coated substrate (Fig. 4, B and D). Over a 210-min period, the two types of aggregate disseminated in a similar manner on laminin-1 and vitronectin. Thereafter, cad7-29 aggregates dispersed more rapidly than Ncad-1 aggregates on vitronectin. In contrast, they did not disperse on type I collagen during this period. On FN, a significant difference in scattering was observed between the two types of aggregate, both in the timing of dispersion and its extent. cad7-29 cells started to escape from the aggregate within 30 min, whereas it took >90 min for Ncad-1 cells to emerge. This difference in dispersion increased rapidly with time and after 5 h the ratio O(t = x)/O(t = 0) for dispersing cad7-29 aggregates was three times higher than that obtained for Ncad-1.

We assessed the ability of cell aggregates to disperse in a three-dimensional environment by analyzing their behavior inside an ECM gel. No invasion was observed within 5 h (data not shown). However, after 24 h both types of aggregate invaded the ECM gel (Fig. 3, E and J) with cad7-29 cells invading more efficiently. In this environment, cad7-29 cells appeared to be connected with only a few other cells and formed aligned cell structures migrating into the ECM gel. Ncad-1 cells did escape from the aggregates, but as strongly cohesive sheets of cells.

One explanation for the differences in dispersion behavior of cad7-29 and Ncad-1 cell aggregates is that these cells have a different pattern of integrins at their surface. We explored this possibility by immunoprecipitation experiments with specific antibodies against integrins; both transfectants had the same integrin pattern as that previously described for parental S180 cells (Beauvais et al., 1995; large amounts of α3β1, αvβ1, and αvβ3 integrins, small amounts of α5β1 and α6β1). A slight increase was found in the amount of α6β1 integrin for Ncad-1 cells (not shown).

Time-lapse videomicroscopy was carried out to analyze the behavior of Ncad-1 (Fig. 5 A) and cad7-29 (Fig. 5 C) cells migrating out of aggregates and to determine their speed of locomotion. The migration of cells was followed until aggregates deposited on FN were totally dispersed. Ncad-1 cells migrating as an epithelial sheet were still strongly apposed to their neighbors after 4 h (Fig. 5 B) and longer (not shown). cad7-29 cells were found in contact with their neighbors at the start of their migratory process but they then began to dissociate and established only transient contacts with their neighbors (Fig. 5 D). They migrated faster than Ncad-1 cells, even when they started to migrate when they were still closely interacting. Locomotion speed was calculated as described in Materials and Methods. We found that cad7-29 cells migrated three times faster than Ncad-1 cells (57.2 ± 14.4 and 19.5 ± 4.7 μm/h, respectively). As the two transfectants displayed similar levels of proliferation (data not shown), their speed of locomotion probably reflects the major difference in their area ratio, calculated after 5 h for aggregate dispersion. It also suggests that, despite high cell density favoring cell-cell contacts inside the outgrowth area, migrating cad7-29 cells gradually became less able to adhere to other cells on FN.

Parental S180 cells migrated on FN with an intermediate speed of locomotion of 37.4 ± 12.8 μm/h, but this was measured for isolated cells because S180 cells does not form aggregate. One possible explanation for the speed of the cad7-29 migration on FN being higher than that of S180 cells is the fact that cad7-29 cells are initially aggregated (Fig. 5 A). A population pressure effect of cad7-29 cells emerging from the aggregate on the cells already migrating on FN could increase the motility of these cells. We analyzed the effect of population pressure on S180 cells by Varani assay. S180, Ncad-1, and cad7-29 were artificially maintained in an aggregated form (independently of cell adhesion molecules) inside an agarose droplet and deposited on FN substrate. The droplets were cultured for 20 h to allow the cells to escape and migrate from this artificial aggregate. In these conditions, we obtained larger outgrowth areas for cad7-29 cells (Fig. 5 E) than for S180 (Fig. 5 F) or Ncad-1 (Fig. 5 G) cells. Quantification was performed for at least 24 droplets of each cell type (Fig. 5 H) and confirmed the results obtained in videomicroscopy experiments. Thus, in two independent assays we observed that cad7-29 cells migrated faster than S180 and Ncad-1 cells.

Another explanation for these differences in migratory properties is that cad7-29, Ncad-1, and parental S180 cells differentially adhere to or spread on FN. We performed a standard cell-FN adhesion assay in which isolated cells were exposed to various concentrations of FN-coated substrates. We observed that all three cell types adhered and spread in a similar manner to substrates coated with 0.5–50 μg/ml of FN (not shown). We then performed experiments with lower concentrations of FN (0.01–0.5 μg/ml). Cell adhesion to FN (Fig. 6 A) occurred between 0.01 and 0.05 μg/ml. S180 cells initially seemed to adhere more strongly than cad7-29 and Ncad-1 cells at a concentration of 0.05 μg/ml of FN. However, for higher FN concentrations, there was no obvious difference between the three cell types; we obtained almost 80% cell adhesion at 0.1 μg/ml with maximum of cell adhesion reached at 0.5 μg/ml. Cell spreading (Fig. 6 B) was calculated by measuring the area of adherent cells for each FN concentration tested. We found that cad7-29 (Fig. 6 D) and Ncad-1 (Fig. 6 E) cells were always more flattened than S180 (Fig. 6 C) cells. This was more evident at a concentration of 0.5 μg/ml which gave maximum cell adhesion.

The prevalence of the proteins involved in cell adhesion during aggregate dispersion was analyzed by immunola-beling with specific antibodies directed against N-cadherin, cadherin-7, β-catenin, or vinculin and with phalloidin-FITC which interacts specifically with filamentous actin. Cell aggregates were deposited on glass coverslips with or without a coating of FN and were allowed to scatter for several hours before fixation.

All Ncad-1 cells escaping from aggregates on both uncoated or FN-coated glass coverslips were closely connected to their neighbors. However, Ncad-1 cells appeared to spread more extensively on FN. N-Cadherin (Fig. 7, A and B, top panel), β-catenin (Fig. 7 C, top panel), and α-catenin (not shown) were strongly detected at cell-cell boundaries. F-Actin was organized as a cortical network along cell-cell contacts and in numerous filaments anchoring both at the tips of lamellipodia (Fig. 7 D, white arrow-head, top panel) or at cell-cell contacts (adherens junctions; Fig. 7 D, black arrow, top panel). This shows that the locations of N-cadherin–mediated cell-cell contacts are not affected by adhesion to FN.

Cadherin-7 was also strongly located at cell-cell contacts in cad7-29 cells migrating out of aggregates cultured on glass coverslips (Fig. 7 A, bottom panel). On FN, cad7-29 cells were more flattened and less connected to neighboring cells. The extent of cell-cell contacts for cad7-29 cells and the level of the expression of cadherin (Fig. 7 B, bottom panel), β-catenin (Fig. 7 C, bottom panel), and α-catenin (not shown) at these sites were lower than those for Ncad-1 cells. Some cells close together produced small amounts of cadherin-7 at cell boundaries. The overall level of cadherin-7 expression appeared lower than that of cells cultured on uncoated coverslips. F-Actin was less organized in cad7-29 cells than in Ncad-1 cells, especially at cell-cell contact sites, where it was found only as a cortical network (Fig. 7 D, black arrow, bottom panel). Fewer actin filaments than in Ncad-1 cells were anchored at cell-cell contacts. Stress fibers were detected at the tips of lamellipodia (Fig. 7 D, white arrowhead, bottom panel) and as small dots along cad7-29 cell-FN contacts. The distribution of vinculin was similar in the two cell types, detected as small dots corresponding to focal sites (not shown). Thus, cadherin-7–mediated cell-cell contacts were regulated differently from those mediated by N-cadherin when cells were interacting with FN.

We further analyzed whether the reduction in the ability of cad7-29 cells to establish cell-cell contacts on FN was reversible. We plated cad7-29 aggregates on this substrate and allowed them to disseminate. We then perturbed cell-FN interactions by incubating cells in the presence of RGDS peptide, inhibitory antibody directed against mouse β1 chain of integrins, or control IgG. We used a concentration of competitors insufficient for complete cell detachment during the assay. Within 1 h of treatment, changes in cell morphology were clearly observed for cad7-29 cells on FN (Fig. 8, top panel). The flattened cell shape observed in the presence of control IgG (Fig. 8 A) was changed in the presence of competitors such as RGDS peptides or anti-β1 antibodies (Fig. 8, C and D). Cells were more rounded due to a higher level of cell-cell adhesion. Large amounts of cadherin-7 were present at intercellular contacts. We investigated whether this was due to a redistribution of the protein or to the incorporation of newly synthesized pools of cadherin-7 by treating cells with CHX at a final concentration of 10 μg/ml for 1 h before incubation of the cells with or without competitors for another hour. Under these conditions, we observed a slightly smaller expression of cadherin-7 if the cells were cultured on FN (Fig. 8 B). RGDS peptides (Fig. 8 F) and to a lesser extent anti-β1 antibodies (Fig. 8 E) interfered with cell-FN interactions in CHX-treated cells and an increase in cadherin-7 expression at cell-cell contacts was observed. Thus, migrating cad7-29 cells seem to possess a pool of cadherin-7 that can be rapidly mobilized and incorporated into newly formed cell-cell contacts.

For Ncad-1 cells (Fig. 8, bottom panel), anti-β1 antibodies caused a decrease in cell spreading which correlated with larger amounts of N-cadherin at cell-cell contacts (Fig. 8 C) than those in untreated cells (Fig. 8 A). A similar result was obtained with RGDS peptides (not shown). Treatment with CHX (Fig. 8 B) gave no significant decrease in the overall level of N-cadherin expressed by the cells. Antibodies against the integrin β1 chain caused an increase in N-cadherin levels at cell-cell contacts (Fig. 8 D). These data suggest that, as for cadherin-7, there may be a pool of N-cadherin that is rapidly mobilized and incorporated into cell-cell contacts when cell-FN interactions are inhibited. However, N-cadherin levels seem to be less affected than those of cadherin-7 if protein synthesis is inhibited. Western blot analyses of β-catenin, cadherin-7, and N-cadherin levels were carried out with cad7-29 and Ncad-1 extracts. Cell extracts were prepared from untreated cells and from cells treated with 10 μg/ml of CHX for various periods of time (Fig. 9). β-Catenin and cadherin-7 were more rapidly downregulated in cad7-29 cells treated with CHX than in Ncad-1 cells. After 5 h of CHX treatment, β-catenin and cadherin-7 levels in cad7-29 cells had decreased, whereas β-catenin levels were stable and N-cadherin levels were only slightly lower after 5 h in CHX-treated Ncad-1 cells. Thus, the two cadherins seem to have different protein turnover rates and cadherin-7–containing cell-cell contacts have higher turnover rates than N-cadherin–containing cell-cell contacts. This correlated with our videomicroscopy observations that cad7-29 cells established only transient cell-cell contacts, whereas Ncad-1 cells established durable cell-cell contacts. Protein synthesis was required for cad7-29 cells to establish these transients contacts and to maintain the overall level of cadherin-7 during cell migration.

Isolated cad7-29 and Ncad-1 cells had similar adhesion and spreading properties on FN (see Fig. 6) but had different spreading, scattering, and migratory properties when escaping from aggregates. We analyzed whether N-cadherin– or cadherin-7–mediated cell adhesion differentially affected downstream integrin-signaling molecules. We focused our study on FAK and its phosphorylation because this protein is one of the principal downstream elements of integrin signaling and is involved in the control of cell adhesion and motility (Guand and Shalloway, 1992; Cary et al., 1996; Gilmore and Romer, 1996; Ilic et al., 1997). We performed Western blots to compare tyrosine-phosphorylated proteins in Ncad-1 and cad7-29 cells engaged in cell-cell and/or cell-FN interactions (Fig. 10). Some proteins in cells plated on FN (lane 2) were specifically tyrosine-phosphorylated, as shown by comparison with cells maintained in suspension (lane 1) or plated on BSA (lane 4). The most highly phosphorylated of these proteins is FAK (arrows, top panels), as previously described by other laboratories. Cells maintained in suspension for 2 h are engaged in strong aggregation process (almost 75%, see also Fig. 2). In this case, one protein with a molecular mass of up to 130 kD appeared to be specifically phosphorylated (Fig. 10, arrowhead). This band was also detected if cells had initiated their cell-cell adhesion process and were plated on BSA, whereas it was not detected in cells plated on FN. Other proteins were found identically phosphorylated in all conditions tested. If Ncad-1 or cad7-29 cells were allowed to initiate cadherin-mediated cell-cell adhesion before plating on FN, there was a decrease in the tyrosine phosphorylation of FAK (lane 3 versus lane 2). Two Western blots were performed and both gave similar results. The blots were stripped and incubated with anti-FAK antibody to determine the amount of FAK in each extract. We then compared, by scanning densitometry, the ratio of phosphorylated FAK to total FAK (Fig. 10, bottom panel) corresponding to the FAK phosphorylation index for untreated cells or cells treated to initiate cell-cell adhesion before plating on FN. N-Cadherin–mediated adhesion had a stronger effect than cadherin-7–mediated adhesion on FN-dependent FAK phosphorylation. Thus, the differences in behavior of aggregated Ncad-1 and cad7-29 cells on FN may be due to different effects of N-cadherin–and cadherin-7–mediated adhesion on integrin-dependent signaling, thereby regulating the initial response of the cells to FN.

Small Ncad-1 and cad7-29 aggregates were grafted into embryos to assess their behavior in vivo. They were inserted into the NCC migratory pathway, under the ectoderm and between neural tube and somite at the axial level, where NCC were migrating along their ventral pathway (Beauvais et al., 1995). The distribution of cells was established following fixation of the host embryos 18 or 48 h after receiving the graft. For S180 cells, which do not aggregate, we grafted a small piece of a cell pellet obtained by centrifugation.

18 h after the graft, parental S180 cell behavior was as previously described (Beauvais et al., 1995). These cells dispersed from the pellet and migrated as individual cells in the interface between the dermomyotome and sclerotome. They also migrated more ventrally and reached the notochord environment (not shown). After 18 h, cad7-29 cells still formed an aggregate at or close to the graft site (Fig. 11 A, arrowheads). NCC invaded the cad7-29 aggregate, as shown by specific NC-1 immunostaining (Fig. 11, A and B, arrows). This suggests that cad7-29 cells become loosely connected within the aggregate, permitting the penetration of NCC, probably by cadherin-7–mediated interactions. Ncad-1 cell aggregates do not disperse at all into the embryo. They remain as strongly cohesive aggregates at the graft site and NCC do not penetrate the aggregate (not shown). This suggests that cadherin-mediated adhesion functions in vivo and that migratory NCC interact transiently with cad7-29, but not with Ncad-1 cells.

Embryos killed 48 h after the graft were used to quantify the aggregate dispersion patterns. We calculated the percentage of embryos with cells located in arbitrarily defined dorsoventral regions (Fig. 12, D and E). Cells actively migrating inside embryonic structures (for example in Fig. 12, A and B) were often found scattered within two consecutive levels. In such case, the score in Fig. 12 E corresponds to the maximum level reached by cells. We also investigated whether cad7-29 cells were present at specific sites within the embryo, when compared with parental cells. During these additional 30 h in the embryo, parental S180 cells continued to migrate ventrally. More than 50% of embryos analyzed had cells that had reached levels 3 and 4 and some cells were found closer to the aorta (Fig. 12 A, arrow). cad7-29 cell aggregates were dispersed and cells migrated ventrally, mainly as individual cells (Fig. 12 B, arrow). All embryos analyzed had cells scattered in levels 2 and 3. In almost 90% of embryos injected with Ncad-1 aggregates, there were Ncad-1 cells at or close to the graft site (Fig. 12 C, arrow). Migrating cad7-29 cells in levels 2 and 3 did not incorporate into dorsal root ganglia or associate with dorsal and ventral roots, which produce both N-cadherin and cadherin-7 (Duband et al., 1988; Nakagawa and Takeichi, 1998). Some cad7-29 cells were detected in close contact with ventral roots (Fig. 13, A and B, arrowheads) or dorsal root ganglia but this was also observed for parental S180 cells (Fig. 13, C and D, arrowheads). These results suggest that in vivo cad7-29 cells cannot be specifically arrested in cadherin-7–expressing tissue.

N-Cadherin and cadherin-7 are functional in transfected S180 cells. The strong and similar upregulation of β-catenin suggests that Ncad-1 and cad7-29 cells produce similar amounts of cadherins and that their different behaviors in vitro and in vivo are not primarily due to differences in cadherin synthesis.

Ncad-1 cells migrated less than cad7-29 cells on FN, which suggests that N-cadherin mediated more stable cell-cell contacts than cadherin-7. This notion was supported by the observation that the two cadherin types seemed to have different turnover rates, with cadherin-7–based cell-cell contact turnover being higher than that of N-cadherin–mediated cell-cell contacts. Furthermore, cad7-29 migratory cells established transient contacts on FN which may be a major event in the control cell motility by inducing locomotion, as previously suggested (Thomas and Yamada, 1992). Mass cell migrations may require a weak intercellular adhesion system to produce directionality. Differences in the forces generated by cadherin subtype-mediated adhesion may also regulate cell migratory properties. This remains to be demonstrated but our observation that the initial rate of adhesion was lower for cad7-29 than for Ncad-1 cells is consistent with this hypothesis.

N-Cadherin may also directly suppress cell motility, whereas cadherin-7 may not. It has been shown recently for E-cadherin that a portion of the cytoplasmic domain, the juxtamembrane domain, plays a role in the suppression of cell motility (Chen et al., 1997). In addition, if Ncad-1 cells are engaged in cell adhesion before plating on FN, there is a greater reduction in the FN-dependent phosphorylation of FAK than with cad7-29 cells. This is consistent with FAK being involved in the control of cell motility and may account for the fact that Ncad-1 cells escaping from aggregates are less motile and scattered than cad7-29 cells on FN.

Recently, the molecular basis of cross-talk between inte-grins and cadherins has been investigated by local stimulation of cell surface cadherins with FN- or cadherin-coated beads. Positive long-range autoregulation of N-cadherin–mediated adhesion has been observed (Levenberg et al., 1998). Conversely, if integrins are stimulated with FN-coated beads, there is a slight reduction in N-cadherin–mediated adhesion in CHO cells (Levenberg et al., 1998). It would be very interesting to test whether this type of FN-dependent signal is more efficient at destabilizing cadherin-7–mediated rather than N-cadherin–mediated adhesions in the cells used here. The perturbation of cad7-29 cell interactions with FN by competitors caused a large increase of cell adhesion and cadherin-7 expression at contact sites. This suggests that the molecular control of cadherin-7–mediated adhesion is also partly dependent upon FN-dependent signaling events, as previously described for N-cadherin expression on migratory NCC (Monier-Gavelle and Duband, 1995, 1997). FN is the more efficient of the diverse ECM components tested for discriminating between Ncad-1 and cad7-29 behavior, which suggests that the molecular control is specific to FN-dependent signaling. The long-range autoregulation observed in CHO cells may also occur in Ncad-1 cells to maintain their strong cohesion. It may not occur in cadherin-7–expressing cells but this issue remains to be investigated.

The two types of aggregate also behave differently and in a more drastic manner in vivo than in vitro. Ncad-1 cells did not disperse in vivo. Even after 48 h they remained aggregated at the graft site. 18 h after the graft, parental S180 cells were migrating into embryonic pathways as NCC (although in some aspects they still exhibited distinct features; Beauvais et al., 1995), whereas cad7-29 cells were only found close to the graft site. This result seems to conflict partly with our in vitro data. However, as S180 cells could not self-aggregate, they were injected as a piece of cell pellet, which could not mimic cell aggregate. This would explain why S180 cells start to migrate as soon as they are injected. No disruption of cell-cell adhesion is required before migration, in contrast to cad7-29 cells. After 18 h, cad7-29 cells seemed to form loosely connected aggregate, suggesting that cell-cell adhesion was being disrupted; mixing was also observed between these cells and NCC, probably mediated by cadherin-7. This phenomenon was not observed for Ncad-1 cells: the aggregate is too cohesive and NCC do not synthesize N-cadherin when they are migrating.

Later, cad7-29 aggregates continued to disperse and the cells started to migrate into embryonic structures. This is consistent with the previous observation that cadherin-7 mRNA is present in migratory NCC and some of their derivatives (Nakagawa and Takeichi, 1995). It has been suggested that cadherin-7 promotes cell sorting in the NCC population and is involved in their targeting. In this study, we observed that cad7-29 cells interacted transiently with endogenous NCC before migrating ventrally into embryonic structures but were not specifically arrested to the vicinity of ventral roots or dorsal root ganglia (structures expressing cadherin-7). Thus, cadherin-7 may be more crucial for the migration than for NCC targeting to specific sites, especially if the ECM environment contains a large amount of FN. It has been shown that FN and type I collagen are particularly abundant along the migratory pathways of NCC, whereas laminin and type IV collagen are found in the basal surfaces of epithelia but are rarely associated with NCC. The different effects of FN and type I collagen on in vitro cad7-29 cell aggregate dispersion suggest that FN rather than type I collagen is the principal mediator of cad7-29 cell motility in vivo.

Large amounts of cadherin-7 are expressed in dorsal root ganglia raising the question of its role in tissue formation. Cadherin-7 may well have a role in the maintenance of cell-cell adhesion between neuronal cells because FN levels are very low in the ganglia (Thiery et al., 1982). However, N-cadherin is more likely than cadherin-7 to be involved in the aggregation of NCC. Ncad-1 cells cannot disperse in NCC pathways. This inhibition of cell migration in cells that express N-cadherin contrasts with the ability of N-cadherin substrates to promote neuritic outgrowth, although neurite elongation occurs by a different mechanism: new membranes are added at the base of the growth cone with the neurite shaft remaining immobile (Doherty and Walsh, 1991). However, it correlates with the fact that NCC cease to produce N-cadherin before they emigrate from neural tube. In a recent study, Nakagawa and Takeichi (1998) observed ectopic expression of N-cadherin on isolated migratory NCC in the ventral pathway after injecting recombinant adenovirus into the lumen of the neural tube. Nevertheless, as the authors themselves pointed out, the lag time between adenovirus infection and protein synthesis may have induced N-cadherin synthesis at a late stage of NCC migration.

Our observations that cadherin-7–mediated cell adhesion is regulated by the ECM environment and that N-cadherin and cadherin-7 adhesion differentially affect FN-dependent signaling indicate another way of regulating cell behavior in vivo. It would be very interesting to investigate the role of other cadherins such as cadherin-11, produced in large amounts by mesenchymal and migratory cells in vitro and in vivo, and to analyze whether behavior on ECM components differs. The analysis of molecules differentially involved in the cross-talk between type I and type II cadherin-mediated adhesion and integrin signaling should provide insight into the molecular mechanisms controlling tissue remodeling and cell migration during early embryogenesis.

possible_ack_statement_here

We thank Professor M. Takeichi and Dr. S. Nakagawa for generously donating cad7-29 and Ncad-1 cells. We also thank Professor M. Takeichi for helpful comments on the manuscript. We are grateful to P.-J. Ridway and Dr. J.-P. Saint-Jeannet for their critical reading of the manuscript. We thank J.-B. Sibarita and D. Morineau for their help in imaging and computerized videomicroscopy.