c-Src-Mediated Epithelial Cell Migration and Invasion Regulated by PDZ Binding Site^{▿ †}

Cell lysis, Wb, and antibodies.

Cells were lysed in modified radioimmunoprecipitation assay buffer containing 150 mM NaCl, 50 mM Tris (pH 7.4), 1 mM EDTA (pH 8.0), 1.0% NP-40, 0.25% Na-deoxycholate, 2 mM Na-vanadate, 25 mM NaF, and a protease inhibitor cocktail (Roche). Western blotting (Wb) was carried out using standard procedures. The antibodies used were mouse monoclonal anti-Src clone GD11 and anti-phospho-Tyr4G10 (Upstate); rabbit polyclonal anti-Src2, anti-focal adhesion kinase c-20, anti-cortactin H-191, and anti-extracellular signal-regulated kinase 2 (anti-ERK2) c-14 (Santa Cruz Biotechnology); anti-phospho-Y861 focal adhesion kinase and anti-phospho-Y421 cortactin (Biosource); anti-phospho-Y416 Src family, anti-phospho-Y527 Src, and anti-phospho-p44/42 mitogen-activated protein kinase (MAPK; Cell Signaling Technology); antigiantin (Covance); anti-MT1-MMP (catalog no. AB8345); and anti-laminin-5 γ2 chain (catalog no. MAB19562) (Chemicon). MPP2 is a custom-made (Eurogentec) rabbit polyclonal antibody raised against a peptide in the C terminus of human MPP2. Integrated pixel densities from Wbs were quantified using NHI image software.

Cell fractionation.

Cells were scraped into sonication buffer (50 mM HEPES [pH 7.4], 135 mM NaCl, 3 mM KCl, 3 mM EDTA, 2 mM Na-vanadate, 25 mM NaF, and a protease inhibitor mini cocktail [Roche]) and sonicated by three 15-s pulses (at 18 W). Nuclei and intact cells were pelleted by centrifugation at 850 × g for 5 min. Supernatant containing 400 μg of protein (postnuclear supernatant) was recovered and centrifuged at 100,000 × g for 20 min at 4°C to obtain a soluble fraction (S100). For Triton extraction, S100 membrane pellets were incubated for 1 h at room temperature in the sonication buffer containing 1.0% Triton X-100 (TX-100). Extracted membranes were centrifuged at 100,000 × g, as described above, to obtain TX-100-soluble (P100 TX-100 sol.) and TX-100-insoluble (P100 TX-100 insol.) fractions. All fractions were solubilized in Laemmli sample buffer. One-twentieth of S100, 1/4 of P100 TX-100 sol., and 1/4 of P100 TX-100 insol. solutions were used for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Wb.

Csk assay.

Hemagglutinin (HA)-tagged Src proteins were produced by coupled in vitro transcription/translation (using a reticulocyte lysate kit; Promega). HA-tagged proteins were immunoprecipitated from reticulocyte lysates and subjected to kinase reaction mixtures with the following components: 50 mM Tris (pH 7.5), 1 mM EGTA, 0.1% 2-mercaptoethanol, 0.1 mM Na₃VO₄, 10 mM MnCl₂, 15 mM MgCl₂, 100 μM ATP, 4 mM morpholinepropanesulfonic acid, 5 mM β-glycerol phosphate, 0.2 mM dithiothreitol with or without 100 ng C-terminal Src kinase (Csk; Upstate) per reaction. Reactions were carried out at 30°C for 20 min.

Immunofluorescence microscopy.

Cells were fixed in 3.6% formaldehyde at room temperature for 15 min or in 100% methanol at −20°C for 2 min, permeabilized with 0.5% TX-100 for 5 min, and blocked with 10% fetal bovine serum in phosphate-buffered saline. Primary antibodies were used at 1/200 dilution in 5% fetal bovine serum in phosphate-buffered saline. Secondary anti-mouse or anti-rabbit antibodies (Jackson ImmunoResearch) coupled to fluorescein isothiocyanate or tetramethylrhodamine isothiocyanate were used for detection. Confocal image stacks were acquired with an inverted Leica SP5 microscope through a Leica 63× oil objective, recorded with a Leica camera, and analyzed with Imaris version 5.3 software. Fluorescence intensities of polarized migrating cells (broad, flat lamella at the front and at trailing edges at the rear) were quantified using NIH image software.

Cell migration assays.

For wound healing assays, confluent cells were starved in starvation medium for 18 h. Wounds were created by scratches with pipette tips. Cells were incubated with or without 20 ng/ml EGF in a temperature-controlled and CO₂-controlled climate chamber mounted onto a Leica inverted microscope. Time-lapse images were captured every 10 min for 15 h using a Hamamatsu ORCA ER C4742-95 camera through a 10× Leica N Plan objective and operated by Openlab software. For single-cell migration, subconfluent cells were detached by trypsinization, recovered in trypsin inhibitor solution (Sigma), and resuspended and kept in suspension in the starvation medium for 1 h at 37°C. MCF-10A cell basement membrane (BM) was prepared as described previously (38). Briefly, MCF-10A cells were grown to confluence on the desired support and treated with 20 mM ammonium hydroxide to destroy the cells while preserving the BM. Cells were seeded onto the BM with or without 20 ng/ml EGF and incubated for 2 h at 37°C before being processed for immunofluorescence (IF) analysis. For single-cell migration analysis by live cell imaging, time-lapse images were captured every 4 min for 2 h as described above. To quantify cell movements, the localization of individual cells within an area of 600 by 600 digital pixels at n time points was recorded with Openlab as x/y values. A 2-h trajectory of the migrating cells was plotted with Excel based on 30 x/y values for each cell. Total distance (d) was determined as . Velocity was quantified as d/t and expressed as μm/min (pixel size was 0.64 μm). Directionality was determined by dividing the direct distance (from the start to the end point) of each cell trajectory by the corresponding d value.

Golgi body reorientation measurements.

The positions of the Golgi apparatus in migrating cells were determined according to the method described by Nobes and Hall (41).

Acinar culture growth in three-dimensional basement membrane cultures.

Lab-TekII chamber slides were coated with 40 μl of Matrigel (BD Biosciences) per well. Cells (10⁴) in 400 μl of complete medium containing 2% Matrigel were seeded per well on top of the solidified Matrigel. Medium was replaced with complete medium containing 2% Matrigel every 4 days.

Gelatin zymography.

Cell extracts in Laemmli buffer without β-mercaptoethanol and without boiling were separated by SDS-PAGE in gels containing 1 mg/ml gelatin. Gels were incubated in 2.5% TX-100 for 30 min, and an in-gel protease cleavage reaction was performed thereafter in a buffer containing 50 mM Tris (pH 8.0), 200 mM NaCl, 5 mM CaCl₂, 0.02% Brij35 at 37°C overnight. Gelatin cleavage was revealed by Coomassie staining.

Deregulated Src-A activity is independent of Csk and can be rescued by an alternative PDZ ligand.

Negative regulation of Src family kinases involves phosphorylation of Tyr527 (pY527) by Csk (8). Conserved residues in the kinase domain of Src family kinases specify the recognition of Src family kinases as substrates by Csk. All these residues are located N-terminal of Tyr527, indicating that the C-terminal GENL sequence is not required for Csk phosphorylation of c-Src (32, 50). However, to exclude the possibility that Src-A cannot be phosphorylated by Csk and hence displayed increased activity, we compared phosphorylation on Tyr527 of Src-L and Src-A. We expressed Src-L and Src-A or the kinase domains of Src-L or Src-A (SrcCT-L and SrcCT-A, respectively) in vitro and used them as substrates for purified active Csk (Fig. ​(Fig.22 A). Phosphorylation of Tyr527 in Src was detected using anti-pY527 Src antibody. The addition of active Csk led to a substantial increase in pTyr527 in all Src proteins tested compared to the reactions carried out in the absence of Csk, and we detected no differences between Src-L and Src-A. To determine whether Src-A is an equally good substrate for Csk in cells, we probed total cell lysates of Src-L and Src-A cells with anti-pY527 Src antibody. No differences in pTyr527 were detectable, whereas Src-A phosphorylation of Tyr416 was increased (Fig. ​(Fig.2B2B).

Open in a separate window

FIG. 2.

Deregulated Src-A activity is independent of Csk and can be rescued by an alternative PDZ ligand. (A) HA-tagged full-length Src-L and Src-A or truncated (the kinase domain including the C terminus) SrcCT-L and SrcCT-A were expressed in reticulocyte lysates. Anti-HA immunoprecipitates were subjected to a kinase reaction with or without Csk. Tyr527 Src phosphorylation was detected by Wb. Expression controls for the individual proteins are shown below. (B) Wbs of the total cell lysates of Src-L and Src-A cells grown in complete medium using anti-pY416 Src, anti-pY527 Src, and anti-Src antibodies. (C) Src-L (GENL), Src-A (GENA), Src-STEV (STEV), and Src-STEA (STEA) were expressed in HEK293 cells. Wbs are shown of total cell lysates of cells grown in starvation medium using the antibodies indicated to the right of each panel. Cortactin was immunoprecipitated, and immunoprecipitates were subjected to anti-pTyr Wb.

We next determined whether the deregulated Src activity promoted by mutating the PDZ ligand sequence GENL to GENA could be rescued by replacing GENA with an artificial PDZ ligand sequence. To this end, we replaced GENA at the c-Src C terminus with STEV, a high-affinity binder for AF6 (44). Expression of Src-GENA in HEK293 cells increased the phosphorylation of Y416 Src, of total Tyr, and of cortactin (Fig. ​(Fig.2C).2C). Expression of Src-STEV led to kinase activity and phosphorylation patterns comparable to those of cells expressing Src-GENL. Conversely, expression of Src-STEA, a mutation that does not bind AF6, displayed kinase activity and phosphorylation patterns similar to those of Src-GENA and reminiscent of deregulated Src activity. We detected no differences in Y527 Src phosphorylation between Src with an intact PDZ ligand sequence and Src with a mutated PDZ ligand sequence. Taken together, these data show that the Src activity and function toward endogenous substrates can be regulated by modifying the specificity of the Src C terminus for PDZ proteins by a mechanism that is independent of Csk phosphorylation of Src.

Src-A expression disrupts the morphology of acinar cultures.

Overexpression of Src-A in mouse fibroblasts that do not express Src family kinases Src, Yes, and Fyn (SYF^−/− cells) promotes focus formation, anchorage-independent growth, and invasion (44). Therefore, we investigated whether Src-A interfered with morphogenetic processes or led to increased proliferation in nontransformed epithelial cells in three-dimensional cultures. MCF-10A cells seeded onto extracellular matrix grow into spheroids, named hereafter acinar cultures, and are characterized by a polarized alignment of cells and apoptosis of central cells that do not contact the BM (16).

Eleven-day acinar cultures derived from Src-A cells, but not from MCF-10A or Src-L cells, showed clusters of protruding cells (Fig. ​(Fig.3A,3A, arrows). We stained the BM of acinar cultures with laminin-5 (Ln-5)-specific antibody and nuclei with Hoechst compound to determine the integrity of the BM by IF microscopy (Fig. ​(Fig.3B).3B). The Ln-5 staining in all MCF-10A and Src-L acinar cultures enclosed all cells and appeared circular, whereas it was interrupted in some (approximately 25%) Src-A cultures and resembled a crescent moon with the opening toward the side of the protruding cell clusters. Neither in acinar nor in two-dimensional cultures did we observe increased proliferation of Src-A cells, suggesting that Src-A interfered with cell polarization rather than with cell proliferation to produce this phenotype.

Open in a separate window

FIG. 3.

Src-A expression in MCF-10A cells disrupts acinar cultures. (A) Bright-field images of 11-day acinar cultures of MCF-10A, Src-L, and Src-A cells. Arrows indicate clusters of cells that protrude from the original acinar culture. (B) IF microscopy images, as in panel A, with anti-Ln-5 antibody (red). Nuclei stained with Hoechst compound are in blue. Bars correspond to 20 μm.

Src-A impairs cell polarization during wound healing.

To determine whether the C-terminal leucine in GENL is critical for cell polarization, we investigated collective cell migration after wounding, which requires cell polarization (17). Collective cell migration and wound healing in starved MCF-10A cells can be induced by EGF stimulation (34), and EGF-stimulated MCF-10A cells were able to close a 0.5-mm-wide scratch within 12 h (Fig. ​(Fig.4A;4A; also see Movies S2 and S4 in the supplemental material). Migration was blocked in the presence of the Src kinase inhibitor PP2, implying that Src family kinase activity is required for efficient cell sheet migration (Fig. ​(Fig.4A;4A; also see Movie S3 in the supplemental material). Src-L cells closed the wound within 12 h after EGF stimulation (Fig. ​(Fig.4B;4B; also see Movie S5 in the supplemental material), whereas Src-A cells failed to close the wound within 12 h, with marginal cells displaying random rather than directed movements (Fig. ​(Fig.4B;4B; also see Movie S6 in the supplemental material). Trajectories of migrating MCF-10A cells at the wound margin aligned parallel to an overall direction toward the wound (Fig. ​(Fig.4B).4B). Trajectories of Src-L cells appeared more heterogeneous during the initial phase (at <4 h) but later aligned in a parallel direction (at >4 h). Src-A cells failed to align in parallel trajectories, with some cells even moving in opposing directions.

Open in a separate window

FIG. 4.

Src-A expression impairs cell polarization. (A) Wound-healing assay. Where indicated, dimethyl sulfoxide (DMSO) solvent or 10 μM of Src kinase inhibitor PP2 was added. Wound healing was monitored for 12 h by live video microscopy. Still pictures are shown from video microscopy at 0 and 12 h (see Movies S1 to S3 in the supplemental material). (B) The same still images shown in panel A comparing the MCF-10A, Src-L, and Src-A cells (see Movies S4 to S6 in the supplemental material). Cells at the leading edge at 0 h were tracked for 12 h and plotted. Each line represents the path of one single cell over time. Axes are in digital pixels (0.64 μm/dp). (C) The Golgi apparatus was visualized by using anti-giantin antibody, and the position of the Golgi body relative to that of the nucleus was determined. Approximately 100 cells from five different fields were analyzed for each cell line, and the percentage of cells with correctly oriented Golgi bodies was determined. Bars show percentages of correctly oriented Golgi bodies and corresponding standard errors of the means. (D) Ln-5 (red), the Golgi apparatus (green), and the nuclei (blue) were visualized in 11-day acinar cultures by confocal IF microscopy. Bars correspond to 20 μm.

The failure of Src-A cells to establish directional persistence could result from impaired cell polarization toward the wound. In migrating, polarized cells, the Golgi apparatus is localized between the nucleus and the leading edge (6). We quantified Golgi apparatus orientation at 6 h after EGF-induced collective cell migration (Fig. ​(Fig.4C).4C). Data revealed that 63% of the MCF-10A and 61% of the Src-L cells analyzed showed Golgi body localization between the leading edge and the nucleus. In contrast, Src-A expression led to a significant decrease of cell polarization because only 44% of Src-A cells displayed Golgi apparatus orientation toward the wound. Hence, the attenuated wound-directed migration of Src-A cells is in part the consequence of impaired polarization toward the wound.

To investigate whether the impaired polarization of Src-A cells could also contribute to the disruption of acinar cultures, we visualized Ln-5-positive BM and Golgi bodies in 11-day acinar cultures (Fig. ​(Fig.4D).4D). Ln-5-positive BM delimited MCF-10A and Src-L acinar cultures. The Golgi apparatus of MCF-10A or Src-L cells localized on the side of the nucleus toward the center of the acinus, away from the BM, and indicated proper apicobasal polarization (49). In Src-A acinar cultures displaying a crescent moon-shaped Ln-5 staining pattern, cells escaped into the surrounding matrix with Golgi bodies oriented toward the presumable direction of migration. Taken together, Src-A expression impairs marginal cell polarization during EGF-induced collective cell sheet migration and apicobasal cell polarization in acinar cultures.

Src-A impairs localization of E-cadherin.

The failure of Src-A cell polarization could be the result of a defect in cell-cell contact organization. Key players in epithelial cell-cell and cell-matrix adhesion are the adhesion molecule E-cadherin and its intracellular linker β-catenin (37). During wound healing, E-cadherin localized to regions of cell-cell contact up to the foremost rows of MCF-10A and Src-L cells (Fig. ​(Fig.5A,5A, upper row). Src-A cells showed decreased and punctuated E-cadherin staining in regions of cell-cell contact. In complete growth medium, confluent monolayers of control cells displayed homogenous E-cadherin staining in regions of cell-cell contact, whereas the Src-A cell shape was irregular, and most cells displayed frayed E-cadherin staining (Fig. ​(Fig.5A,5A, middle row). Analogous to that of E-cadherin, β-catenin was distributed in linear cell-cell contacts of MCF-10A and Src-L cells, whereas in Src-A cells, β-catenin was distributed along ridged, filamentous structures (Fig. ​(Fig.5A,5A, lower row). These data suggest that GENL-regulated c-Src activity and localization (44) are required for the maintenance of cell sheet integrity and the proper E-cadherin and β-catenin localization in MCF-10A cells.

Open in a separate window

FIG. 5.

Src-A expression impairs localization of E-cadherin. (A) Localization of E-cadherin during wound healing (8 h after wounding; top row) and in monolayers (middle row) or of β-catenin in monolayers (bottom row) was visualized by IF microscopy. For better visualization, IF staining is shown in the inverted mode. (B) Eleven-day acinar cultures of the MCF-10A, Src-L, and Src-A cells. Black and white images show inverted confocal IF of E-cadherin distribution. Color images show corresponding acinar cultures stained for E-cadherin (red), membrane protein palmitoylated 2 (green), and nuclei (blue). Arrows indicate the boundary between the acinar cultures and the ECM. Fourfold magnification of the framed area shows an escaping cell (arrowheads) with intact nucleus (n). (C) Confocal IF image of the Src-A acinar culture stained with anti-E-cadherin antibody. Arrowheads indicate punctuated E-cadherin staining in the escaping cell (fourfold magnification of the framed area). Bars correspond to 20 μm, unless otherwise stated.

We next investigated E-cadherin localization with 11-day acinar cultures of MCF-10A, Src-L, and Src-A cells, using confocal IF microscopy. E-cadherin staining was detectable in regions of cell-cell contact (Fig. ​(Fig.5B).5B). Interestingly, E-cadherin staining at the periphery of Src-A acinar cultures was decreased. Moreover, in Src-A acinar cultures, we found cells that protruded over the acinar margins (Fig. ​(Fig.5B).5B). Cells protruding over the acinar margins displayed decreased E-cadherin staining (Fig. ​(Fig.5B)5B) or E-cadherin staining in punctuated patterns (Fig. ​(Fig.5C,5C, arrowheads), which might be indicative of increased E-cadherin endocytosis (5) induced by Src-A in peripheral regions in contact with BM proteins.

Src-A increases the velocity and directional persistence of cells migrating on BM.

To determine whether Src-A cells respond differently to contact with the BM, we investigated adhesion to and single-cell migration on BM. Ln-5 is enriched in the BM of MCF-10A cells (38) and can be visualized with the Ln-5 γ2 chain-specific antibody (Fig. ​(Fig.8).8). Src-L or Src-A cells seeded onto prepared BM adhered and spread within 60 min (Fig. ​(Fig.6A),6A), lamellipodia formed (Fig. ​(Fig.6A,6A, arrowheads), and cells acquired a migratory phenotype. In the presence of the Src kinase inhibitor PP2, cells were still able to adhere to BM but no lamellipodia formed. We compared cell spreading by quantifying the area covered by individual Src-L and Src-A cells at 60 min after cells were seeded onto BM. We found that the average area covered by individual Src-A cells was twice the size of the area covered by Src-L cells (Fig. ​(Fig.6A,6A, bar diagram). Inhibition of lamellipodium formation by PP2 indicated that Src kinase activity was necessary and implied Src kinase activation during adhesion. We tested this possibility by seeding Src-L and Src-A cells onto standard tissue culture plastic and determined the phosphorylation of Y416 Src (Fig. ​(Fig.6B).6B). Src-L pY416 increased slightly within the first 5 min before returning to basal levels between 30 and 60 min later. Src-A pY416 increased similarly; however, the decline was less marked. To determine whether the BM might provide additional clues to Src activation, we monitored the phosphorylation of Y416 of Src-L and Src-A in cells seeded onto the BM (Fig. ​(Fig.6C).6C). Src-L pY416 declined after 120 min, whereas Src-A pY416 remained elevated. Thus, our data suggest a negative regulatory role of GENL for Src kinase activation after adhesion.

Open in a separate window

FIG. 6.

Src-A increases velocity and directional persistence of migration on BM. (A) Src-L or Src-A cells seeded onto BM of MCF-10A cells in the presence of dimethyl sulfoxide (DMSO) solvent or of 10 μM PP2. Images were recorded 1 h after seeding. Arrowheads indicate lamellipodia in solvent-treated cells. The area covered by individual cells was quantified using NIH image software. Bars show averages of areas covered and standard deviations (n = 19 cells from four randomly selected spots). (B and C) Src-L and Src-A cells were trypsinized and kept in suspension in starvation medium for 1 h at 37°C, seeded onto regular tissue culture plastic (B) or onto the BM of MCF-10A cells (C), and lysed at the indicated time points after seeding. Wb of total cell lysates, using anti-Src2 and anti-pY416 Src antibodies. (D) MCF-10A, Src-L, or Src-A cells were seeded as described in the legend to panel C. Cells were left unstimulated or treated with EGF. Cell movements were recorded by live cell video microscopy over 2 h, tracked, and plotted. (E) The average velocity and median directionality of cells treated as described in the legend to panel D were determined. Average velocities ± standard errors of the means (SEM) were as follows: MCF-10A cells, 0.96 ± 0.043 μm/min without EGF and 1.42 ± 0.1 with EGF; Src-L cells, 1.2 ± 0.10 μm/min without EGF and 1.64 ± 0.11 μm/min with EGF; Src-A cells, 1.45 ± 0.1 μm/min without EGF and 1.95 ± 0.24 μm/min with EGF. Medians ± SEM of directionality were as follows: MCF-10A cells, 0.31 ± 0.026 without EGF and 0.64 ± 0.067 with EGF; Src-L cells, 0.46 ± 0.081 without EGF and 0.55 ± 0.033 with EGF; Src-A cells, 0.67 ± 0.057 without EGF and 0.69 ± 0.046 with EGF (n = 45 randomly selected cells from four experiments; y-axis error bars are SEM).

Open in a separate window

FIG. 8.

Increased Ln-5 processing in Src-A cells. (A) MT1-MMP (green) and nuclei (blue) were visualized by IF microscopy of nonpermeabilized Src-L or Src-A cells at 3 h after cells were seeded onto BM in starvation medium. (B) The MCF-10A, Src-L, and Src-A cells were seeded, and Ln-5 was visualized as described in the legend to panel A (Ln-5). Grayscale Ln-5 fluorescence was inverted and overlaid onto bright-field images (overlay). Areas with decreased Ln-5 are in red. (C) Schema of human Ln-5 γ2 chain with the putative MT1-MMP cleavage site within laminin-type EGF-like subdomains of domain III (according to rat Ln-5 γ2 chain [46]). (D) Wbs of the Ln-5 γ2 p155 and p105 chains in total MCF-10A, Src-L, and Src-A cell extracts using anti-Ln-5 antibody. Expression controls for Src and actin are shown. Pixel densities of the p155 and p105 bands from three independent experiments were quantified, and the ratio of p105/p155 was determined (error bars are standard errors of the means). (E) Gelatin zymography using supernatants of the MCF-10A, Src-L, and Src-A cells grown for 18 h in complete or starvation medium. (F) Gelatin zymography and Wb using membrane extracts of the MCF-10A, Src-L, and Src-A cells grown for 18 h in complete medium. SDS-PAGE was performed under nondenaturing conditions. (G) Ln-5 (red), MT1-MMP (green), and nuclei (blue) were visualized in 11-day acinar cultures by IF confocal microscopy. Bars correspond to 20 μm, unless otherwise stated.

We next determined by live cell video microscopy whether Src-A expression altered the motility of single cells on the BM. We used either starvation medium or starvation medium supplemented with 20 ng/ml EGF. We tracked individual cells (Fig. ​(Fig.6D)6D) and quantified their velocity and directionality (Fig. ​(Fig.6E).6E). The average velocity of MCF-10A cells was 0.96 μm/min without EGF and increased to 1.42 μm/min with EGF. Src-L velocity was 1.2 μm/min without EGF and 1.64 μm/min with EGF. Src-A velocity was 1.45 μm/min without EGF and 1.95 μm/min with EGF. The median directionality of MCF-10A cells was 0.31 without EGF and increased to 0.64 with EGF. Src-L cell directionality was 0.46 without EGF and 0.55 with EGF. Src-A cell directionality was 0.67 without EGF and 0.69 with EGF. These data show that Src-A expression increased velocity and directionality of single cells migrating on the BM in the absence of EGF stimulation to levels similar to that of EGF-stimulated control cells. In contrast, confluent Src-A cells failed to migrate directionally in monolayers during wound healing in the absence and presence of EGF (Fig. ​(Fig.4).4). The combination of cell sheet and single-cell migration analyses indicated that the C-terminal GENL sequence plays a regulatory role by modulating Src kinase activity and function downstream of cell-cell and cell-matrix adhesion. Hence, we suggest that intact c-Src is required to correctly integrate signals stemming from adhesion receptors into pathways regulating MCF-10A cell migration and that GENL is critically implicated in this process.

Src-A promotes phosphorylation of cortactin and FAK and their accumulation near the leading edge.

The c-Src substrates cortactin and FAK are key regulators of cell migration (11). We therefore tested whether Src-A increased cortactin and FAK phosphorylation in motile cells on the c-Src-specific sites Y421 and Y861, respectively. We scratched multiple wounds into a subconfluent layer of cells to induce cell motility and compared the activation status of c-Src, Src-L, and Src-A and of cortactin and FAK after 12 h without and with EGF stimulation (Fig. ​(Fig.7A).7A). The overall activities of c-Src, Src-L, and Src-A investigated by pY416 Src detection in scratched cells at low density (Fig. ​(Fig.7A)7A) did not differ markedly from that of nonscratched cells grown at high density (Fig. ​(Fig.1D),1D), and Y861 FAK phosphorylation levels were similar (Fig. ​(Fig.7A).7A). In contrast, phospho-Y421 cortactin was detectable in scratched Src-A cells grown at low density (Fig. ​(Fig.7A)7A) but not in nonscratched cells grown at high density (Fig. ​(Fig.1D).1D). One possible explanation is that Src-A promotes translocation of cortactin and makes it accessible for phosphorylation. We therefore investigated cortactin localization in motile cells. We seeded MCF-10A, Src-L, and Src-A cells in starvation medium onto BM and visualized cortactin after 2 h by using standard IF (Fig. ​(Fig.7B).7B). We quantified the fluorescence intensity of cortactin staining from the back to the front in cells that displayed a clearly polarized phenotype. Cortactin accumulated toward the leading edge (Fig. ​(Fig.7B,7B, arrows) of Src-A cells but to a much lesser extent in Src-L or MCF-10A cells (Fig. ​(Fig.7B,7B, bar diagrams). Together, these data show that Src-A increased cortactin Y421 phosphorylation and promoted cortactin localization toward the leading edge, possibly contributing to the increased persistence of migration (11) that we observed (Fig. ​(Fig.6E).6E). FAK is required for the spatial organization of the leading edge of migrating cells (48), and FAK is a Src substrate in MCF-10A cells (Fig. ​(Fig.11 and ​and7).7). We therefore investigated whether Src-A increased pY861 FAK toward the leading edge, using confocal IF microscopy. As for the cortactin localization shown in Fig. ​Fig.7B,7B, we seeded cells in starvation medium onto BM for 2 h. Phospho-Y861 FAK was detectable in migrating MCF-10A, Src-L, and Src-A cells primarily at the basal side of lamellipodia and to a lesser extent in the cell body (Fig. ​(Fig.7C,7C, arrowheads). Compared to MCF-10A cells, total pY861 FAK increased in Src-L cells and more markedly in Src-A cells (Fig. ​(Fig.7C,7C, bar diagram). Hence, analogous to cortactin, the increased pY861 FAK shown in Fig. ​Fig.7A7A is likely occurring in the lamella toward the leading edge.

Open in a separate window

FIG. 7.

Src-A promotes phosphorylation of cortactin and FAK and leads to their accumulation toward the leading edge. (A) Wounds (160) were scratched into 50% confluent monolayers of starved MCF-10A, Src-L, and Src-A cells. Cells were then incubated without (−) or with (+) EGF for 12 h. Wb of total cell lysates, using the antibodies indicated, is shown. (B) MCF-10A, Src-L, or Src-A cells were seeded in starvation medium onto BM. Two hours after cells were seeded, cortactin was visualized by IF microscopy. Average fluorescence intensities of one-pixel-wide sections across clearly polarized cells (n = 6 polarized cells per cell line) were determined by using NIH image software and plotted. The orientation is trailing-to-leading edge; the dark gray bars are standard error of the mean (SEM) densities of grayscale fluorescence. Arrows indicate approximate positions of peak fluorescence near leading edges; arrowheads indicate the centers of gravity (peak fluorescence near the nucleus). (C) The MCF-10A, Src-L, and Src-A cells were seeded onto the BM as described in the legend to panel B. Phosphorylation of Y861 FAK was determined by confocal IF microscopy. Arrowheads indicate Y861 FAK phosphorylation at leading edges. The bar diagram shows the averages of integrated pixel densities per cell of 15 cells. Error bars are SEM.

We thank Matthias Hoechli and the electron- and light microscopy facility (EMZ) for excellent support for live cell imaging and confocal microscopy.

This work was supported in part by the Swiss National Science Foundation (K. Moelling).