Scrib is required for maintenance of an epithelial phenotype at low cell densities

When cells were transfected with pS-ScrbKD vectors and plated at low densities, they consistently displayed a mesenchymal phenotype. Normal MDCK cells organize into discrete, tight islands with smooth boundaries, but cells lacking Scrib appeared more fibroblastic (Fig. 2, A and B). They spread over a much larger surface area (approximately three to five times larger; Fig. 2 C), and the edges of the cell clusters were disorganized as though the cells were moving apart from one another. When stained with phalloidin, the normal cortical actin rings were absent in cells lacking SCRIB and were replaced by stress fibers often oriented along the long axis of the cells (Fig. 2 B). These observations suggested that Scrib might regulate epithelial cell adhesion and/or migration.

Open in a separate window

Figure 2.

Cells lacking Scrib lose their epithelial morphology at low density. (A) Control (Luc) and ScrbKD cells were grown on 6-well plates at low density for 3 d. Images were obtained by phase-contrast microscopy using a 10× objective. (B) Control and ScrbKD cells were fixed and stained with phalloidin to visualize F-actin. Typical colonies are shown. (C) Surface areas per cell (μm²) were measured for 80–90 frames (as shown in A) and sorted into bins 200 μm² wide. The percentage of cells in each bin is shown.

Scrib inhibits cell motility and is required for oriented migration

To determine whether Scrib regulates MDCK cell motility, transfected cells were plated onto 8-μm filters in Boyden chambers and incubated in normal medium containing 10% serum (both above and below the filter). The same number of cells was plated onto each filter. After 16–20 h, cells that had migrated through the pores to the bottom surface of the filters were stained with crystal violet. Loss of Scrib substantially increased the number of cells that had migrated through the filter (Fig. 3, A and B). Importantly, when we expressed a GFP fusion of a human Scrib in the cells transfected with the pS-ScrbKD1 vector, the number of cells migrating through the filter was reduced to control levels (Fig. 3, A and B). It was conceivable that the reversion to a wild-type phenotype was caused by reexpression of the endogenous protein. To test for this possibility, we blotted cell lysates for Scrib (Fig. 3 C). The endogenous Scrib protein level was reduced upon expression of the ScrbKD1 shRNA and did not respond to coexpression of the human GFP–Scrib fusion. The GFP–Scrib fusion can be distinguished by its lower mobility on SDS-PAGE and was expressed at approximately two to three times the level of the endogenous protein in control cells (Fig. 3 C). These data prove that the effects of Scrib RNAi on motility are indeed caused by loss of the Scrib protein rather than by off-target effects of the shRNA.

Open in a separate window

Figure 3.

Suppression of Scrib expression increases cell motility and attenuates orientated migration. (A) Motility of cells transfected with Luc, ScrbKD1 shRNA, or ScrbKD1 plus GFP-tagged human Scrib was measured using a Boyden chamber assay. Images were captured after crystal violet staining of the cells at the bottom side of the filter. (B) Quantification of cell migration by measuring A595 of eluted crystal violet. Error bars represent mean ± SD. (C) Immunoblot analysis of transfected cells with anti-Scrib antibody. Equal protein concentrations were loaded and normalized using anti-tubulin. (D) Rose plot of individual cell tracks from time-lapse movies of the wounding assay.

The filter assay depends not only on cell motility but also on the rate of cell attachment to and spreading on the filter surface. Therefore, as an alternative approach to measuring cell motility, we performed wounding assays on MDCK monolayers and tracked the movement of individual cells within the population at the wound edge by time-lapse microscopy. The overall rate of wound closure was similar for both the control cells and those lacking Scrib (0.35 vs. 0.32 μm/min). However, the behavior of cells lacking Scrib was remarkably different from that of the control (Videos 1 and 2, available at http://www.jcb.org/cgi/content/full/jcb.200506094/DC1). At the edges of the wound, control cells extruded lamellipodia and moved forward as an organized sheet, but the KD cells were less organized. Some of the KD cells lost their attachment to the sheet, pulled away from the leading edge, and moved in random directions. This difference can be seen in the Rose plots of cells tracked over the period of the assay (Fig. 3 D). Calculation of directionality parameters confirmed that the cells lacking Scrib move at a significantly higher speed than control cells (2.54 ± 0.19 vs. 1.73 ± 0.41 μm/min; P = 0.026) and with a lower persistence coefficient (4.3 vs. 11.6 min; P = 0.02). The reduced persistence accounts for the similarity in the overall rate of wound closure. Interestingly, cells further back from the wound exhibited a continual jiggling motion, as if they had lost adhesion to their neighbors and were trying to move away from one another, and transient gaps appeared in the monolayer (Video 2). These results support the data shown in Fig. 2, suggesting that loss of Scrib causes a defect in cell–cell adhesion.

Scrib effects are not mediated by βPIX binding

Cell movement is regulated by Rac, and this GTPase has also been implicated in controlling adhesion between epithelial cells (Ehrlich et al., 2002; Van Aelst and Symons, 2002; Chu et al., 2004). Interestingly, Scrib has been reported to bind, via its PDZ domains, to the COOH terminus of the Rac guanine nucleotide exchange factor, βPIX (Audebert et al., 2004). To determine whether the SCRIB–βPIX interaction is important in mediating the regulation of cell migration and adhesion, we first confirmed that in MDCK cells we could detect this interaction (unpublished data) and then assessed the role of βPIX by silencing expression of the canine protein in MDCK cells.

Of four pSUPER constructs tested, three efficiently suppressed βPIX expression (Fig. 4 A). In particular, the PixKD1 shRNA reduced expression of the protein by >90%. However, loss of βPIX had no detectable effect on cell migration as measured using the Boyden chamber assay (Fig. 4 B). Moreover, in our hands, overexpression of βPIX did not increase cell migration in the filter assay (unpublished data). We then performed double KD experiments in which the expression of both Scrib and βPIX was suppressed (Fig. 4 C). We reasoned that one function of Scrib might be to sequester and inactivate βPIX. In this case, loss of Scrib would release the βPIX, leading to inappropriate activation of Rac and increased migration. If this hypothesis were correct, a double KD would reverse the migration phenotype by removing the excess free βPIX from the cell. As depicted in Fig. 3 (A and B), migration through filters was increased by Scrib KD. However, the coordinate loss of βPIX did not significantly perturb this effect (Fig. 4 B). Note that cotransfection of the PixKD shRNA did not interfere with gene silencing of Scrib (Fig. 4 C). We therefore conclude that Scrib function in cell adhesion and migration is independent of βPIX binding.

Open in a separate window

Figure 4.

βPIX is not involved in Scrib function at adherens junctions. (A) MDCK II cells were transiently transfected with either pS-Luciferase control vector (Luc) or three pSUPER constructs targeting different sequences of canine βPIX mRNA (PixKD). Proteins were analyzed by SDS-PAGE and immunoblot. (B) Quantification of Boyden chamber migration assay with Luc, ScrbKD, and ScrbKD plus PixKD cells. Error bars represent mean ± SD. (C) Immunoblot showing Scrib single KD and Scrib and βPIX double KD. (D) Rac pull-down assays with control and ScrbKD cells in both normal medium (top) and low-calcium medium (bottom). (E) Immunoblot analysis of phospho-ERK in control and ScrbKD cells with and without stimulation by 15 ng/ml HGF.

Because βPIX is a guanine nucleotide exchange factor for Rac, we also asked whether loss of Scrib would alter Rac activity. Rac-GTP was detected by pull-down assays using a GST fusion of the Rac binding domain of PAK. No consistent differences in Rac-GTP were detected, however, in control cells versus those lacking Scrib (Fig. 4 D). When the cells were subjected to a calcium switch, Rac was activated within 2 h of calcium addition in both the control and Scrib KD cells (Fig. 4 E). We therefore conclude that the polarity defects associated with suppression of Scrib expression are independent of βPIX and are not mediated through the Rac GTPase. Finally, we asked whether Scrib might regulate the extracellular signal–regulated kinase (ERK) signaling pathway, which is activated by scatter factor (hepatocyte growth factor [HGF]; Tanimura et al., 1998). HGF induces an EMT in which cells lose adhesiveness and become more migratory, a phenotype similar to that observed in cells lacking SCRIB. However, no significant differences in phospho-ERK were detected when Scrib expression was knocked down either before or after addition of HGF (Fig. 4 E). These data suggest that Scrib does not function to regulate the HGF signaling pathway.

Scrib is required for E-cadherin–mediated adhesion

To determine whether cell–cell adhesion is compromised in the absence of Scrib, we first used an aggregation assay. Cells were trypsinized, triturated to break up clumps into individual cells, resuspended in fresh medium in a hanging drop beneath the lid of a tissue culture plate, and incubated for 18–20 h. Cell aggregation was then assessed microscopically. A dramatic loss of aggregation was apparent in cells expressing either pS-ScrbKD1 or 2 vectors, as compared with the control cells that were transfected with pS-Luc (Fig. 5, A and C). This effect was not a result of differential loss of E-cadherin in the cell suspensions as assessed by immunoblotting lysates from the suspended cell cultures (Fig. 5 B). Importantly, coexpression of human GFP-Scrib reversed the adhesion defect caused by the loss of endogenous SCRIB, proving that the effect of the shRNAs on adhesion is specifically mediated through destruction of the Scrib mRNA rather than through off-target effects (Fig. 5, A and C). We also tested for a possible role for βPIX on aggregation, using an shRNA directed against the canine gene (Fig. 4, A and B). However, loss of βPIX from the cells had no effect on the aggregation of control cells and did not reverse the loss of aggregation observed in the absence of Scrib (Fig. 5 D).

Open in a separate window

Figure 5.

Decreased adhesiveness of cells lacking Scrib. (A) 3 × 10⁴ control and ScrbKD cells were seeded into hanging drop cultures and allowed to aggregate overnight. After trituration by passing the cell cluster 10 times through a 200-μl pipette tip, images were captured by phase-contrast microscopy using a 10× objective. (B) E-cadherin is not preferentially lost from the ScrbKD cells in suspension culture. (C) Quantification of the degree of aggregation shown in A. Data are presented as the area of the aggregated cells/number of individual nonaggregated cells and represent means of 10–12 images from triplicates of each sample ± SD. (D) Effect of βPIX silencing on cell aggregation.

To determine whether the aggregation defect is mediated through E-cadherin or some other cell adhesion protein, we assayed the ability of the MDCK cells to attach to a surface coated with the extracellular domain of E-cadherin. Cells were disassociated using an EGTA solution (with no trypsin), centrifuged and resuspended in fresh medium, and added to 96-well plates coated with the ectodomain of E-cadherin. After 60 min, the plates were washed and remaining attached cells were counted. Results are shown in Fig. 6. Almost no cells attached to the plates in the absence of the E-cadherin ectodomain, demonstrating that during the 60-min incubation period integrin-mediated attachment is negligible. Control cells attached efficiently, and attachment was proportional to the amount of E-cadherin ectodomain on the plate (Fig. 6 B). Importantly, loss of Scrib caused a substantial drop (approximately threefold) in cell attachment, demonstrating that E-cadherin homophilic adhesion is compromised in the absence of Scrib. Addition of an arginine–glycine–aspartic acid peptide to block integrin-mediated adhesion had no significant effect (unpublished data).

Open in a separate window

Figure 6.

Suppression of Scrib expression reduces E-cadherin–mediated adhesion. (A) In vitro E-cadherin binding assay. Different concentrations of recombinant E-cadherin extracellular domain were coated on plates. Control (Luc) and ScrbKD cells were collected in suspension and allowed to adhere to the plates for 60 min. Images were captured for nonwashed plates (to give total cell numbers), and plates were subjected to washing (to determine attached cell numbers). (B) Quantification of cell attachment to E-cadherin ectodomain. Data represent means of 10–12 images from triplicates of each condition ± SD.

Depletion of Scrib causes a defect in cell–cell adhesive junctions

These data suggest that Scrib is required for normal E-cadherin function at cell–cell junctions. We therefore examined the distribution of E-cadherin and of the Na/K-ATPase, which is a marker for the basolateral membrane. In control cells, these proteins colocalize along the lateral cell boundaries. Scrib KD caused a distinctive phenotype in which the lateral membranes of the cells became disorganized. The membranes appeared less vertical and had convoluted edges (Fig. 7 A). A similar phenotype was observed for β-catenin distribution (Fig. 7 B). However, the total amounts of E-cadherin, β-catenin, and α-catenin expressed in cells depleted of Scrib were the same as the amounts in control cells (Fig. 7 C). Scrib does not, therefore, regulate the expression of these junctional proteins. Moreover, when surface proteins were biotinylated, captured on streptavidin beads, and blotted for E-cadherin, no reproducible difference was observed between the control and Scrib KD cells (Fig. 7 D). These data demonstrate that there is no change in the amount of E-cadherin on the cell surface, and we conclude that Scrib is not involved in controlling the exocytosis or endocytosis of E-cadherin.

Open in a separate window

Figure 7.

Loss of Scrib causes a defect in adherens junction structure. (A) Control and Scrib-depleted cells were plated at subconfluence and allowed to form islands of cells. They were then fixed and stained for E-cadherin (red) and Na/K-ATPase (green). Image stacks were collected using confocal microscopy with 0.95-μm z steps. (B) Control and ScrbKD cells were fixed and stained for Scrib and β-catenin. (C) Immunoblot of junctional proteins in control and ScrbKD cells. Whole cell lysates were blotted for E-cadherin, α- and β-catenin, and α-tubulin. (D) Cells were treated with a nonpermeable biotin linker, sulfo-NHS-SS-biotin, and biotinylated proteins were captured onto streptavidin-agarose and detected by immunoblot. (E) Control and ScrbKD cells plated at low density were lysed with buffer containing 0.5% Triton X-100. After centrifugation, supernatants and pellets were resolved by SDS-PAGE. Distributions of α- and β-catenin were detected with immunoblot. 25% of the soluble and 50% of the insoluble fractions are shown. (F) Quantification by Odyssey image system. Data are presented as the ratio of insoluble to soluble fractions. Error bars represent the SD from four independent experiments.

When adherens junctions form, a fraction of the α- and β-catenin becomes stabilized at the cell cortex, either through clustering of the E-cadherin or perhaps through attachment to actin, and is detergent insoluble. We measured the detergent-insoluble fraction in control and Scrib KD cells and found that in the absence of Scrib the amounts of both α- and β-catenin in this fraction were substantially reduced (Fig. 7, E and F). Together, these results suggest that Scrib is required for the normal stabilization of α- and β-catenin at the cell cortex.

Depletion of E-cadherin phenocopies the effects of Scrib KD

If both the adhesion defect and the increased motility observed in response to Scrib silencing are caused by decreased E-cadherin activity, one would predict that depletion of E-cadherin would produce the same phenotype. We therefore expressed a shRNA targeted against the canine E-cadherin in MDCK cells and achieved a >50% reduction in E-cadherin expression (Fig. 8 A). Interestingly, cells depleted of E-cadherin migrated through filters significantly faster than the control (Fig. 8 B). Moreover, these cells were larger and more fibroblastic in appearance than control cells when plated at low densities (Fig. 8 C), just as observed for the Scrib KD cells. Based on these data, we conclude that both the morphological changes and increased motility in cells depleted of Scrib can be ascribed to a failure of the E-cadherin to form normal trans-adhesive interactions.

Open in a separate window

Figure 8.

Effects of E-cadherin depletion and expression of an E-cadherin–α-catenin fusion protein. (A) Cells were transfected with an shRNA targeted against canine E-cadherin. Lysates were immunoblotted for E-cadherin and Scrib. Equal protein concentrations were loaded and normalized using anti-tubulin. (B) Quantification of the motility of cells transfected with Luc or EcadKD shRNAs using Boyden chamber assay as described in Fig. 3. (C) Morphology of cells depleted of E-cadherin. Control and EcadKD cells were grown on six-well plates at low density for 3 d. Images were obtained by phase-contrast microscopy using a 10× objective. (D) Expression of an Ecad–αcat and Ecad–GFP fusion in MDCK cells. The fusion proteins and endogenous E-cadherin were detected by immunoblot with anti–E-cadherin antibodies. (E) Effects of the Ecad–αcat and Ecad–GFP fusions on motility of cells depleted of Scrib. Migration through filters was quantified as in Fig. 3. (F) Aggregation assay. 3 × 10⁴ cells were seeded into hanging drop cultures and allowed to aggregate for overnight. After trituration with a 200-μl pipet tip, images were captured by phase-contrast microscopy using a 10× objective. (G) Quantification of the aggregation assay. Data are presented as the area of the aggregated cells/number of individual nonaggregated cells per field and represent means of 10–12 fields from triplicates of each sample ± SD.

Cell culture, transfection, and calcium switch

Cell culture, transfection, and calcium switch of MDCK II cells were performed as described previously (Chen and Macara, 2005). Cells (2 × 10⁶) were transiently transfected in suspension by electroporation, using 2.5–18 μg DNA (Amaxa, Inc.). Transfection efficiency was generally >70%. For calcium switch experiments, 3 × 10⁴ MDCK II cells were plated into 8-well Lab-Tek II chambers (Nunc) with normal growth medium. After 40–44 h, the medium was replaced with MEM lacking calcium and supplemented with 2% dialyzed calf serum. After 16–20 h, cells were switched back to normal growth medium.

Immunological methods

For analysis of total cell extracts by immunoblot, cells were scraped directly into Laemmli sample buffer. After SDS-PAGE, proteins were transferred to nitrocellulose and detected by chemiluminescence (Kirkegaard and Perry Laboratories) or, for quantification of the proteins, with the Odyssey Infrared Imaging System (LI-COR Corp.). For immunoblots of the soluble fraction, cells were washed with cold PBS and lysed in lysis buffer (25 mM Hepes, pH 7.4, 150 mM NaCl, 5 mM MgCl₂, 0.5 mM EDTA, 1% Triton X-100, 1 mM DTT, and 1 mM PMSF). After centrifugation (10 min at 14,000 rpm), supernatants were assayed for protein concentration using Bradford reagent and then boiled in Laemmli sample buffer.

Antibodies used were as follows: anti-Scrib (1:150; Santa Cruz Biotechnology, Inc.), anti–β-catenin (1:3,000); anti-βPIX (1:500), anti–E-cadherin (1:3,000), anti-ERK (1:5,000), and anti–α-catenin (1:2,000; BD Biosciences); anti–α-tubulin (1:5,000; Sigma-Aldrich); anti-Rac (1:500, Upstate Biotechnology); anti-ERK and phospho-ERK antibodies (provided by D. Lannigan, University of Virginia, Charlotteseville, VA); and 1 μg/ml anti-myc 9E10 and anti-HA 12CA5. HRP-conjugated goat anti–mouse, mouse anti–goat, or goat anti–rabbit secondary antibodies were used at a dilution of 1:5,000–1:10,000 (Jackson ImmunoResearch Laboratories). For Odyssey detection, Alexa Fluor 680–conjugated goat anti–mouse, mouse anti–goat (Invitrogen), or IRDye 800–conjugated goat anti–rabbit (Rockland, Inc.) secondary antibodies were used. For immunofluorescence, cells were usually fixed in 4% paraformaldehyde, permeabilized with 0.5% Triton X-100 in PBS, and blocked with 5% BSA in PBS for 1 h before incubation with antibodies. For detection of Na/K-ATPase, cells were fixed in methanol/acetone (1:1) at −20°C for 10 min. Primary antibodies were anti-Scrib (1:100) or anti–ZO-1 (1:500), anti-occludin (1:500; Zymed Laboratories), anti–β-catenin, anti–E-cadherin (1:500), and anti–Na/K-ATPase (1:250; Abcam). Anti-gp35 (1:300) was a gift from G. Ojakian (State University of New York Downstate Medical Center, Brooklyn, NY). Alexa Fluor–conjugated secondary antibodies (Invitrogen) were used at a dilution of 1:1:500–1:1,000. Alexa 594–conjugated phalloidin was used at a dilution of 1:50. Cells were mounted in Slowfade (Invitrogen).

Imaging

Epifluorescence images were collected using an inverted microscope (T200; Nikon) with a 60× water-immersion lens (Plan Achromatic, NA 1.2) coupled to a charge-coupled device camera (Orca; Hamamatsu), controlled by Openlab 4.0 software (Improvision). Images were collected at 12-bit depth, 1,024 × 1,280 pixels resolution, and converted to TIFF files. Images were postprocessed in Photoshop 7.0 (Adobe) to increase the grayscale range and to reduce haze using an unsharp mask and were converted to 8-bit depth. In some cases (Fig. 1 C), the black and white values were inverted, and the contrast was further enhanced to emphasize cell junctions. Movies were converted to QuickTime format. Confocal imaging was performed using a microscope (LSM510; Carl Zeiss MicroImaging, Inc.) with a 100× oil-immersion objective (Plan Achromatic, NA 1.3).

Triton X-100 solubility

The Triton X-100 solubility assay was performed as previously described (Tsukamoto and Nigam, 1999; Palacios et al., 2002). After transfection, 1.2 × 10⁵ cells were plated on 6-well plates. Cells were lysed after 3 d in 100 μl CSK-A buffer containing 0.5% Triton X-100 for 15 min on ice. Cell lysates were collected and centrifuged at 14,000 rpm for 10 min. Supernatants and pellets were boiled with sample buffer, and equal volumes were resolved by SDS-PAGE. Proteins were visualized by immunoblotting and quantified with the Odyssey Infrared Imaging System.

Cell surface biotinylation

Cell surface biotinylation was performed as described previously (Le et al., 1999). MDCK cells were grown on filters and incubated with 1.0 mg/ml sulfosuccinimidyl 2-(biotinamido) ethyl-dithioproprionate (sulfo-NHS-SS-biotin) (Pierce Chemical Co.) on both sides of the filter for 1 h. Cells were then washed with quenching reagent (50 mM NH₄Cl in PBS containing 1 mM MgCl₂ and 0.1 mM CaCl₂) for 10 min, followed by further washes in PBS. Cells were then scraped off the filters and lysed in RIPA buffer, and cell lysates were incubated with streptavidin beads (Pierce Chemical Co.) to collect biotinylated proteins. Biotinylated E-cadherin was detected by immunoblot.

Rac activity assays

Assays were performed as described previously using a GST fusion of the Rac binding domain of PAK1 to capture Rac-GTP (Ren et al., 1999; Chen and Macara, 2005).

Aggregation assay

The hanging drop aggregation assay was performed essentially as described previously (Thoreson et al., 2000). Cells were trypsinized in the presence of EDTA, washed twice in PBS, and resuspended at 10⁶ cells per milliliter in normal growth medium. 3 × 10⁴ cells were then suspended as hanging drops from the lid of a 24-well culture dish and allowed to aggregate overnight. The cells were subjected to shear force by passage 10 times through a 200-μl filter pipet tip (Continental Laboratory). Cells were then imaged using a 10× phase-contrast objective. For quantification, in each field, the degree of aggregation was measured as the ratio of the area of the aggregates to the number of individual nonaggregated cells using Openlab software (Improvision).

E-cadherin binding assay

The E-cadherin binding assay was a modified version of one described previously (Yap et al., 1998). Wells in a 96-well ELISA high-binding plate (Corning) were coated with 0–50 μg/ml of E-cadherin extracellular domain–Fc fusion protein (provided by B. Gumbiner, University of Virginia, Charlottesville, VA) in coating buffer (100 mM NaCl, 20 mM Hepes, and 1 mM CaCl₂, pH 7.2) overnight at 4°C. Wells were then blocked with 3% BSA for 2 h, followed by three washes in HBSS+ (containing 1.2 mM Ca; Invitrogen). MDCK cells were dispersed using a cell disassociation solution (Sigma-Aldrich) at 37°C for 30 min and washed twice with normal medium. Cells were then resuspended in HBSS+ with 1 mM additional Ca. About 10⁵ cells were added to wells in the presence of 0.2 mg/ml GRGDTP peptide (Sigma-Aldrich) and allowed to incubate for 60 min. After gentle washing with HBSS+, cells bound to plates were imaged. The number of cells in unwashed wells was counted, and data are presented as the percentage of cells remaining in each washed well, compared with the unwashed control.

Boyden chamber cell migration and wound-healing assays

Modified Boyden chamber assays were performed essentially as described previously (Sander et al., 1998). Approximately 10⁵ cells in DME with 10% serum were seeded in the upper compartment of cell culture–treated Transwell filters (6.5-mm diam and 8-μm pores; Corning). The lower compartment contained DME with 10% serum. After 16–20 h at 37°C, nonmigrating cells in the top chamber were removed with a cotton swab and cells that had migrated to the underside of the filter were fixed with 4% paraformaldehyde and stained with 0.4% crystal violet. For quantification, crystal violet was eluted with 10% acetic acid and the absorbance at 595 nm of eluant was measured. To normalize for variability in cell numbers, an identical volume of cells was seeded into another well, fixed, stained, and eluted, without the cells from the upper well being removed.

For the wounding assay, confluent monolayers grown on Delta T dishes (Fisher Scientific) in DME + 10% serum + 20 mM Hepes, pH 7.4, were scraped with the tip of a microinjection pipette to form a linear wound. After 4 h of recovery, wound closure was recorded using a phase-contrast 20× objective lens at 5-min intervals for 16 h at 37°C. Individual cells were tracked over the course of the movie using Openlab software, transposed to the same initial x,y coordinates (0,0), and displayed as Rose plots. The mean square displacement, d ²(t), was calculated for each cell. Migration parameters were then estimated, assuming the cells migrate as persistent random walkers: d ²(t) = 2S ² P[t − P(1 − e^{− t / P})], where S is the cell speed and P is the directional persistence time. An Excel macro for calculating the correlation functions was provided by R. Horwitz (University of Virginia, Charlottesville, VA).

We thank James Nelson for Ecad–GFP, Rick Horwitz for the cell migration analysis macro, Deborah Lannigan for the ERK antibodies, George Ojakian for anti-gp135, members of the Macara group for helpful suggestions, and the anonymous reviewers for insightful comments. We also thank Kathleen Schwartz, Xiao Chen, and Xuejun Chen from the Gumbiner Laboratory for invaluable help with the E-cadherin binding assay. We thank Anne Spang for critical reading of the manuscript.

This work was supported by grants GM070902 and CA040042 from the National Institutes of Health, Department of Health and Human Services.