p120 catenin regulates lamellipodial dynamics and cell adhesion in cooperation with cortactin

p120 Catenin Is Required for Lamellipodial Persistence.

We first produced p120-knockdown MCF-7 epithelial cells by using stable expression of a construct encoding p120 shRNA and confirmed that, similarly to other cultured epithelial cells lacking p120 (2, 3, 8), these cells demonstrate impaired cadherin-mediated cell–cell junctions [supporting information (SI) Fig. 4].

The lamellipodial activity of quiescent epithelial cells can be induced by a variety of motogenic factors, including neuregulin (heregulin) (26), the ligand for the Erb3 and Erb4 receptors (27). To test whether p120-knockdown in epithelial cells affects lamellipodial activity, we examined the response of serum-starved control and p120-shRNA-expressing MCF-7 cells to neuregulin stimulation. As early as 5–10 min after the addition of neuregulin, the majority of control cells exhibited intensive lamellipodial and ruffling activity at the periphery of the epithelial island (Fig. 1 A and F and SI Movie 1a). However, neuregulin-treated, p120-shRNA-expressing MCF-7 cells formed, at the periphery, only sporadic, discontinuous lamellipodial protrusions (Fig. 1 B and G and SI Movie 1b).

Kymograph analysis (28) revealed (Fig. 1 C and D and Table 1) that lamella extension rates were approximately the same in control MCF-7 cells and cells with p120-knockdown. However, the extension phase in control cells persisted longer than in the cells lacking p120, in which the development of protrusions was frequently interrupted by edge withdrawal (compare Fig. 1 C and D and Table 1). Consequently, maximal lamella extension (double-headed arrows in Fig. 1 C and D) and the net cell edge displacement per hour (Table 1) were significantly lower in the p120-deficient cells, compared with controls.

Similar deficiency of lamellipodia persistence was described in cells after RNAi-induced down-regulation of cortactin, a component of ruffles and lamellipodia and a potent activator of Arp2/3-dependent actin polymerization (29). In MCF-7 cells, using kymograph analysis, we demonstrated that treatment with siRNA directed to cortactin, indeed, led to alteration of lamellipodial activity quantitatively similar to that described above for p120-depleated MCF-7 cells (Fig. 1E, SI Movie 1 c and d, and Table 1). It is noteworthy that, at the same time, overall morphology of cortactin-knockdown cells differed from that of p120-knockdown cells; in particular, cortactin-depleted cells, unlike p120-depleated cells, demonstrated augmented substrate spreading (SI Movie 1d) in agreement with a previous publication (30).

Visualization of actin cytoskeleton in control and p120-depleted MCF-7 cells treated with neuregulin revealed reduction of lamellipodia and stress fibers in cells lacking p120 (compare Fig. 1 F and G). At the same time, these cells still displayed nonpolarized actin-rich “internal protrusions” at the interface between neighboring cells (revealed by deconvolution microscopy; Fig. 1G) and, in some cases, dorsal ruffles (SI Movie 1b).

Although p120-knockdown significantly affects neuregulin-induced lamellipodial activity, the general mechanisms of neuregulin-dependent signal transduction were nevertheless preserved in the absence of p120. Indeed, both control and p120-deficient MCF-7 cells display similar activation of the MAP kinases Erk1/2 and PI3 kinase upon neuregulin treatment (SI Fig. 5 A and B).

Effect of p120-Knockdown on Focal Adhesion Formation and Cell Migration.

Besides destabilization of cell–cell junctions and decrease of lamellipodial persistence, adhesion to the extracellular matrix was also suppressed in p120-deficient cells. Significant reduction of the average projected area of p120-knockdown cells, as compared with their control counterparts, was observed in MCF-7 cells (SI Fig. 6 A Upper). Focal adhesions visualized by antivinculin antibody staining in p120-depleted MCF-7 cells (Fig. 1 H and I and SI Fig. 6 C and D) showed significant decrease in area and fluorescence intensity (SI Fig. 6B), as well as in total number per unit of cell area (SI Fig. 6A Lower), compared with focal adhesions in control cells. Moreover, observations of focal adhesion dynamics using cells transiently transfected with mCherry-vinculin demonstrated that the average rate of focal adhesion formation (number of newly formed adhesions per hour) was 32.7 (min 13 to max 59) in control cells and only 5.0 (min 2 to max 10) in p120-knockdown cells (P = 0.05, Wilcoxon's ranked-sum test). Taking into consideration the total number of focal adhesions in cells of each type, we calculated that the rate of focal adhesion turnover (fraction of focal adhesions, which appeared/disappeared per 1 h) was also somewhat higher in control cells than in p120-knockdown cells [0.32 ± 0.08 vs. 0.20 ± 0.08 (mean± SEM)].

The difference in focal adhesion formation between control and p120-depleted MCF-7 cells was especially spectacular in the situation of neuregulin-mediated stimulation, which caused a dramatic activation of focal adhesion formation in control cells (Fig. 1H). The effect of neuregulin on focal adhesions in p120-shRNA cells was manifested by the appearance of only a few mature focal adhesions at the cell peripheral processes (Fig. 1I).

To verify that the effects of RNAi described above are specific, we prepared a p120 construct bearing triple-point mutation in the 19-bp sequence that served as a target for shRNA-mediated knockdown in our experiments. This mutation conserved the amino acid sequence of the p120 protein, but rendered its expression insensitive to inhibition by the shRNA. Expression of this shRNA-insensitive p120 in the cells stably expressing the p120-shRNA, indeed, restored the ability of these cells to form typical adherens junctions (SI Fig. 7 A and B), as well as prominent actin-rich lamellipodia and focal adhesions (SI Fig. 7 C–F).

Requirement of p120 for cell spreading, migration, and focal adhesion formation was also validated in other cell types, including BALB/c 3T3 cells and B16 F1 melanoma cells expressing the p120-shRNA. We have shown that p120-knockdown inhibits migration speed and spreading efficiency in BALB/c 3T3 cells (SI Fig. 8) and reduces focal adhesion formation and spreading in B16 F1 cells (data not shown).

p120 Colocalizes and Directly Associates with Cortactin.

The involvement of p120 in lamellipodia and focal adhesion regulation suggests that it may interact with some proteins controlling formation of these structures. One of the apparent candidates for such interaction is cortactin, whose knockdown, as shown above, similarly to the p120-knockdown, suppresses lamellipodial persistence (see also ref. 29). To explore the relationships between p120 and cortactin, we investigated a possible interaction between these two proteins.

Immunofluorescence localization studies indicated that, in addition to cell–cell junctions, p120 was enriched in actin-containing ruffles and lamellipodia. Such localization, especially prominent in highly motile cells such as B16 F1 or MDA-MB 231 breast carcinoma (Fig. 2A, SI Movie 2a, and SI Fig. 9 A–D) was also observed in MCF-7 cells and BALB/c 3T3 fibroblasts (Fig. 2 B and C and SI Fig. 9 E and F). We used deconvolution-wide field microscopy and total interference reflection fluorescence microscopy to confirm that p120 is indeed enriched in ruffles and at the leading edge of the lamella close to the substrate (SI Fig. 9 A–D).

We next studied the colocalization of p120 and cortactin. Coexpression of GFP-p120 and RFP-cortactin in B16 F1 mouse melanoma cells revealed dynamic colocalization of the two in membrane ruffles at the leading edges (Fig. 2A, asterisks) and in the halos associated with macropinocytic vesicles (Fig. 2A, arrows; see also SI Movie 2 a–c). This observation is in agreement with previous data showing that cortactin is a component of such structures (31, 32). In addition, colocalization of endogenous p120 and cortactin in lamellipodia of neuregulin-stimulated MCF-7 (Fig. 2B) and BALB/c-3T3 cells spreading on fibronectin (Fig. 2C) was also observed.

To determine whether p120 and cortactin are associated with the same molecular complex, CHO cells were transiently cotransfected with p120 and GFP-cortactin, and the proteins were immunoprecipitated from cell lysate by using antibodies against GFP or p120. As shown in Fig. 2D, a complex containing both p120 and GFP-cortactin could be precipitated by using either antibody. To confirm these results with endogenous proteins, we also demonstrated that the endogenous p120 can be specifically immunoprecipitated from MCF-7 cell lysates with anticortactin antibody (Fig. 2E). In the reciprocal experiment, endogenous cortactin was also pulled down from cell extract by using bacterially expressed GST-p120 (Fig. 2F).

Using FLAG-fusion constructs encoding the N-terminal (amino acids 1–330) or C-terminal (amino acids 350–546) parts of cortactin, we demonstrated that GFP-p120 coimmunoprecipitates with the N-terminal portion of cortactin (Fig. 2G), which contains Arp2/3- and actin-binding sites (33–35), but not with the C-terminal region containing the SH3 domain. We next translated the N-terminal domain of cortactin in vitro by using the TNT kit (Promega, Madison, WI) and showed that the translation product interacts with bacterially expressed GST-p120, but not with GST alone (Fig. 2H), suggesting that the proteins directly interact with each other. This interaction was also observed in the presence of a high concentration of latrunculin B, indicating that the binding was not mediated through residual actin filaments (data not shown).

Finally, we expressed GST-fused p120 and cortactin in bacteria, purified both proteins by affinity chromatography on glutathione-Sepharose column and size-exclusion chromatography, and removed the GST tag from GST-cortactin by proteolysis. Then using these purified proteins (Fig. 2I), we assessed their interaction by incubation of increasing concentrations of GST-p120 with a constant amount of cortactin and pulling down the complex by using glutathione-Sepharose beads (Fig. 2J). Amounts of pulled-down cortactin were quantified densitometrically (Fig. 2J Right). Concentration-dependent pull down of cortactin by p120 shows that these proteins, indeed, directly interact with each other.

p120-Knockdown Leads to the Disappearance of Cortactin from the Leading Edge.

Down-regulation of p120 has a profound effect on cellular distribution of cortactin and its immediate partner, Arp3. In control MCF-7 cells, endogenous cortactin is enriched at the tip of the leading lamella and in cell–cell junctions (Fig. 3A). The levels of leading edge- and junction-associated cortactin were, however, dramatically reduced in cells with down-regulated p120 (Fig. 3B), although total cortactin levels in these cells were not altered, as estimated by Western blot analysis (SI Fig. 4). Exogenous RFP-cortactin in control cells is also recruited to the peripheral cell edge and cell–cell junctions (Fig. 3C), where it colocalizes with Arp3 (Fig. 3E). Cells expressing p120-shRNA failed to form a continuous RFP-cortactin- and Arp3-containing rim at the free cell edges and in cell–cell junctions (Fig. 3 D and F). Instead, both RFP-cortactin and Arp3 were abundant in the cytoplasm of these cells. Of note, the knockdown of cortactin does not abolish the localization of p120. Endogenous p120 was still prominent at the leading edges of cortactin-depleted MCF-7 and MDA-MB 231 cells (SI Fig. 10).

Because lamellipodial activity strongly depends on the function of small GTPase Rac and p120 overexpression was shown to augment Rac activity (19, 20), we checked whether knockdown of p120 affects Rac activity and localization. Pull-down assay (36) did not reveal a significant difference in Rac activity between control and p120-knockdown MCF-7 cells in conditions of serum starvation and neuregulin stimulation (SI Fig. 11 A and B). However, immunofluorescence staining with Rac antibody indicated that enrichment of Rac at the leading edges and cell–cell junctions typical for control MCF-7 cells (SI Fig. 11C) was not observed in the cells lacking p120.

Cell Culture.

MCF-7, MDA-MB 231, BALB/c-3T3, B16 F1, and HCT 116 cells were routinely cultured in DMEM supplemented with antibiotics, glutamine, and 10% heat-inactivated FCS (Gibco, Gaithersburg, MD). Establishment of MCF-7 cells, stably expressing shRNA for p120, as well as a transient cell-transfection procedure are described in SI Materials and Methods. Recombinant human neuregulin (NRG1-beta1 EGF Domain, R&D Systems, Minneapolis, MN) was added to serum-starved MCF-7 cells at a final concentration of 100 ng/ml.

cDNA Constructs.

Mouse p120 cDNA in pRcCMV vector (46) was kindly provided by A. Reynolds (Vanderbilt University, Nashville, TN). The GFP-fused p120 construct in the pEGFP-C1 vector was described previously (19), and cloning of GST-p120 is described in SI Materials and Methods. GFP- and RFP-fused cortactin constructs (31) were kindly provided by M. Kaksonen (University of California, Berkeley, CA), and GST-cortactin construct was kindly provided by A. M. Weaver (Vanderbilt University, Nashville, TN). The full-length N- and C-terminal portions of cortactin cloned into pcDNA3 vector (47) were kindly provided by J. T. Parsons (University of Virginia, Charlottesville, VA). mCherry-C2-vinculin plasmid was kindly provided by J. V. Small (Institute of Molecular Biotechnology, Vienna, Austria).

A 19-base sequence (655gatggttatccaggtggca673) common to all known isoforms of human and mouse p120, but absent in other p120 family members, was selected for the design of p120-shRNA. Details of retroviral plasmid construction are described in SI Materials and Methods. Mutant pEGFP-p120 preserving the same amino acid sequence, but containing three point mutations in the target nucleotide sequence (655gatggttaCccaggCggGa673) and therefore resistant to p120-shRNA, was produced as described in SI Materials and Methods. For cortactin-knockdown, we used anticortactin SMARTpool siRNA obtained from Dharmacon (Lafayette, CO).

Immunochemical and Biochemical Procedures.

Details of immunoprecipitation and Western blot procedures are described in SI Materials and Methods. Activation of PI3-kinase/PKB and MAPK pathways upon neuregulin treatment was analyzed by using monoclonal anti-active MAPK antibodies (DP ERK) and rabbit polyclonal anti-phospho-Ser 473 Akt/PKB antibodies purchased from Sigma–Aldrich Israel (Rehovot, Israel). Total ERK and PKB proteins were detected by using anti-MAPK and anti-Akt/PKB Pan (Sigma–Aldrich Israel), respectively. The Rac-GTP pull-down assay was performed by using a Rac activation assay kit (Cytoskeleton, Denver, CO). For in vitro expression of cortactin, coupled transcription–translation reactions were performed by using rabbit reticulocyte lysates (Promega, Madison, WI) as specified by the manufacturer (Promega). Purification of bacterially expressed GST-p120 and cortactin and in vitro binding assay between these two proteins are described in SI Materials and Methods.

Microscopy, Time Lapses, and Quantification.

Immunofluorescence staining was performed as previously described (19). Antibodies and reagents used are characterized in SI Materials and Methods. For live cell imaging, cells were plated on glass-bottom microwell dishes (MatTek, Ashland, MA). Movies were taken using a DeltaVision microscopy system (Applied Precision, Issaquah, WA). Details of focal adhesion measurements, cell motility quantification, and kymograph analysis of lamellipodial activity (28) are given in SI Materials and Methods. The Olympus (Tokyo, Japan) total interference reflection fluorescence (48) was used to determine the localization of endogenous p120 at the ventral cell surface (see details in SI Materials and Methods).