Caveolin-1 Inhibits Epidermal Growth Factor-stimulated Lamellipod Extension and Cell Migration in Metastatic Mammary Adenocarcinoma Cells (MTLn3)

Caveolin-1 is a principal component of caveolae membranes that may function as a transformation suppressor. For example, the human caveolin-1 gene is localized to a suspected tumor suppressor locus (D7S522; 7q31.1) that is deleted in human cancers, including mammary carcinomas. However, little is known about the role of caveolins in regulating cell movement, a critical parameter in determining metastatic potential. Here, we examine the role of caveolin-1 in cell movement. For this purpose, we employed an established cellular model, MTLn3, a metastatic rat mammary adenocarcinoma cell line. In this system, epidermal growth factor (EGF) stimulation induces rapid lamellipod extension and cell migration. Interestingly, we find that MTLn3 cells fail to express detectable levels of endogenous caveolin-1. To restore caveolin-1 expression in MTLn3 cells efficiently, we employed an inducible adenoviral gene delivery system to achieve tightly controlled expression of caveolin-1. We show here that caveolin-1 expression in MTLn3 cells inhibits EGF-stimulated lamellipod extension and cell migration and blocks their anchorage-independent growth. Under these conditions, EGF-induced activation of the p42/44 mitogen-activated protein kinase cascade is also blunted. Our results suggest that caveolin-1 expression in motile MTLn3 cells induces a non-motile phenotype.

Caveolin-1 is a principal component of caveolae membranes that may function as a transformation suppressor. For example, the human caveolin-1 gene is localized to a suspected tumor suppressor locus (D7S522; 7q31.1) that is deleted in human cancers, including mammary carcinomas. However, little is known about the role of caveolins in regulating cell movement, a critical parameter in determining metastatic potential. Here, we examine the role of caveolin-1 in cell movement. For this purpose, we employed an established cellular model, MTLn3, a metastatic rat mammary adenocarcinoma cell line. In this system, epidermal growth factor (EGF) stimulation induces rapid lamellipod extension and cell migration. Interestingly, we find that MTLn3 cells fail to express detectable levels of endogenous caveolin-1. To restore caveolin-1 expression in MTLn3 cells efficiently, we employed an inducible adenoviral gene delivery system to achieve tightly controlled expression of caveolin-1. We show here that caveolin-1 expression in MTLn3 cells inhibits EGF-stimulated lamellipod extension and cell migration and blocks their anchorage-independent growth. Under these conditions, EGF-induced activation of the p42/44 mitogen-activated protein kinase cascade is also blunted. Our results suggest that caveolin-1 expression in motile MTLn3 cells induces a non-motile phenotype.
Caveolin-1, a 21-24-kDa integral membrane protein, was the first protein shown to be a marker for the membrane microdomains known as caveolae (1)(2)(3). Caveolae are ϳ 50 -100-nm vesicular invaginations of the plasma membrane and are thought to form as a result of a local accumulation of cholesterol, glyco-sphingolipids, and caveolin-1 (4 -6). Two other members of the caveolin gene family have been described, termed caveolin-2 and -3 (7,8). Caveolin-2 has the same tissue distribution as and co-localizes with caveolin-1, whereas caveolin-3 is found only in cardiac and skeletal muscle cells (9,10).
Although the functions attributed to caveolae and the caveolin gene family are still a matter of intense research, their role in vesicular and cholesterol trafficking and their ability to modulate signal transduction events have been studied widely (for review, see Refs. 4 -6). Biochemical and morphological experiments have shown that a variety of signaling molecules are concentrated in these plasma membrane microdomains. This is particularly true of lipid-modified signaling molecules, such as Src family tyrosine kinases, H-Ras, eNOS, and heterotrimeric G-proteins (11)(12)(13)(14)(15)(16). Furthermore, caveolin-1 plays an inhibitory role in many of these signaling events by interacting directly with the membrane-bound components of the involved pathways or their downstream elements.
Because a number of signaling molecules inhibited by caveolin-1 are involved in cell growth and mitogenesis (e.g. Src family tyrosine kinases, the epidermal growth factor (EGF) 1 receptor kinase, Neu tyrosine kinase, Ras, components of the p42/44 mitogen-activated protein (MAP) kinase cascade) (11,12,(17)(18)(19), caveolin-1 has been proposed to function as a putative tumor suppressor. In support of these observations, the human caveolin-1 gene is localized to a suspected tumor suppressor locus (D7S522; 7q31.1) that is deleted in a number of human cancers, including mammary carcinomas (5, 20 -22). In addition, caveolin-1 is modified or down-regulated in cells transformed by activated oncogene products. For example, caveolin-1 was first described as a transformation-dependent substrate of v-Src in Rous sarcoma virus transformed fibroblasts (1). In addition, caveolin-1 levels are reduced dramatically in H-Ras (G12V)-and v-Abl-transformed NIH 3T3 cells (23), and mutational activation of c-Neu down-regulates caveolin-1 expression (18). Antisense-mediated reduction of caveolin-1 levels in normal NIH 3T3 cells leads to hyperactivation of p42/44 MAP kinase cascade, anchorage-independent growth in soft agar, and tumor formation in nude mice (24). Thus, caveolin-1 expression may somehow regulate anchorage-dependent cell growth, contact inhibition, and/or migratory potential by acting as an inhibitor of the p42/44 MAP kinase cascade. This hypothesis remains to be tested.
Motile MTLn3 cells were first described as a spontaneously occurring lung metastasis clonally derived from 13762 rat mammary adenocarcinoma (25,26). EGF stimulation of MTLn3 cell motility has been used previously as a model to study the regulation of lamellipod extension and cell migration. EGF acts as a chemoattractant in a number of cultured cells, and EGF along with its cognate receptor (EGF-R) enhances the growth, migration, and invasion of a number of cancers (27)(28)(29)(30)(31). In addition, up-regulation of EGF-R often signals a poor cancer prognosis (31). Segall and colleagues (32)(33)(34) have used the MTLn3 cell line extensively to study the effects of EGF on lamellipod extension and chemotaxis, demonstrating that actin polymerization at the leading edge of lamellipodia plays a significant role in their extension and subsequent cellular motility.
Here, we examine the potential role of caveolin-1 in the regulation of cell movement by employing the well established MTLn3 cellular model. We find that MTLn3 cells fail to express detectable levels of endogenous caveolin-1, whereas the MTC cell line (the non-motile counterpart of MTLn3 cells) expresses significant levels of caveolin-1. In addition, we show that adenovirus-mediated gene transfer of the caveolin-1 cDNA to MTLn3 cells is sufficient to inhibit EGF-stimulated lamellipod extension and cell migration and to block the anchorage-independent growth of these cells in soft agar. Our results suggest that restoration of caveolin-1 expression in motile MTLn3 cells can induce a non-motile phenotype.

EXPERIMENTAL PROCEDURES
Materials-The caveolin-1 mouse mAb 2297 and caveolin-2 mouse mAb 65 (used for immunoblotting (9,35)) were the gifts of Dr. Roberto Campos-Gonzalez, Transduction Laboratories, Inc. The caveolin-1 rabbit pAb N-20 (used for immunoblotting) and the anti-Myc mouse mAb 9E10 (used for immunofluorescence) were purchased from Santa Cruz Biotechnology. Rhodamine-conjugated phalloidin (used for immunofluorescence) was purchased from Sigma. MTLn3 and MTC cell lines (kindly provided by Dr. J. E. Segall, Albert Einstein College of Medicine, Bronx, NY) were clonally derived from the parental 13762 rat mammary adenocarcinoma (25,36). Normal human mammary epithelial cells were obtained from BioWhittaker/Clonetics, and the human mammary carcinoma cell lines MCF-7 (HTB-22) and T-47D (HTB-133) cells were obtained from ATCC. All other biochemicals used were of the highest purity available and were obtained from regular commercial sources. DNA manipulations, including ligations, bacterial transformation, and plasmid purification, were carried out using standard procedures.
Cell Culture-MTLn3 and MTC cells were grown in alpha minimal essential medium (␣-MEM) supplemented with 5% fetal calf serum and antibiotics. NIH 3T3 cells and Madin-Darby canine kidney cells were grown in Dulbecco's modified Eagle's medium (high glucose) containing 10% donor bovine serum and antibiotics. Normal human mammary epithelial cells and the human mammary carcinoma cell lines (MCF-7 and T-47D) were maintained as described by the suppliers (Clonetics and ATCC, respectively). All cells were passaged at ϳ 60 -80% confluence.
Construction of Recombinant Caveolin Adenoviral Vectors-The adenoviral vector (pAd-tet) used consisted of a tetracycline-regulatable expression cassette (a heptamer of tetO sequences preceding a CMV immediate early promoter). The full-length cDNA for caveolin-1 (canine) was amplified by polymerase chain reaction with a c-Myc epitope tag fused to the COOH terminus (35) and subcloned into the transfer vector pAd-tet using SalI (5Ј) and NotI (3Ј) restriction sites. The correct orientation and sequence of the insert were verified by restriction mapping and DNA sequencing. Human embryonic kidney 293 cells containing the E1 early gene region of the adenovirus type 5 genome were subsequently used to package the adenovirus by co-transfection of 293 cells with the SpnI restriction fragment of pAD-tet-caveolin-1 (pAdcav-1) and the large right end fragment of the Ad5/DE1DE3 genome (adenovirus type 5 genome lacking the E1 and E3 early gene regions). Positive plaques were purified and expanded in 293 cells, and virus titers were determined by A 260 and plaque assay as described previously (37). The adenoviral vectors containing tetracycline-regulatable green fluorescent protein (Ad-GFP) and the CMV-driven tet-controlled transactivator (Ad-tTA) were constructed similarly.
Evaluation of Caveolin-1 Expression Driven by the Viral Vectors-A Madin-Darby canine kidney cell line, which stably expresses tTA, was used to assess the feasibility of adenovirus mediated caveolin-1 gene transfer (38). Cells grown in six-well plates at a density of 4 ϫ 10 5 cells/well were incubated with Ad-cav-1 with an m.o.i. of 50 -200 plaque-forming units/cell in serum-free medium for 1 h, followed by growth in complete medium for 2 days. Cell lysates were prepared and subjected to immunoblotting (see below). For the expression of caveolin-1 by Ad-cav-1 in MTLn3 cells, subconfluent cells were co-infected with Ad-cav-1 together with Ad-tTA (transactivator) for 1 h in serumfree ␣-MEM and grown in complete medium for 1-2 days. Cells were then lysed and subjected to immunoblotting with anti-caveolin-1 IgG. Various m.o.i. values of Ad-cav-1 and Ad-tTA were employed to achieve optimal levels of caveolin-1 expression. To determine the infection efficiency, MTLn3 cells were grown on coverslips in a 24-well plate at 4 ϫ 10 4 cells/well and co-infected with Ad-cav-1 and Ad-tTA at m.o.i. values of 50 -300. One day post-infection, the cells were processed for immunofluorescence microscopy. As a control, Ad-GFP was used to co-infect MTLn3 cells together with Ad-tTA, and the optimal m.o.i. values for Ad-GFP were also determined. To achieve more than an ϳ80 -90% infection rate, we determined the optimal m.o.i. values for Ad-cav-1 plus Ad-tTA or Ad-GFP plus Ad-tTA to be ϳ100 for each. These m.o.i. values were used for infection in subsequent experiments, unless indicated otherwise.
Immunoblot Analysis-Cells, cultured in their respective media, were washed with PBS and incubated with lysis buffer (10 mM Tris, pH 7.5, 50 mM NaCl, 1% Triton X-100, and 60 mM octyl glucoside) containing protease inhibitors (Roche Molecular Biochemicals). Protein concentrations were quantitated using the BCA reagent (Pierce), and the volume required for 10 g of protein was determined. Samples were separated by SDS-PAGE (12.5% acrylamide) and transferred to nitrocellulose. The nitrocellulose membranes were stained with Ponceau S (to visualize protein bands) followed by immunoblot analysis. All subsequent wash buffers contained 10 mM Tris, pH 8.0, 150 mM NaCl, and 0.05% Tween 20, which was supplemented with 1% bovine serum albumin and 2% non-fat dry milk (Carnation) for the blocking solution and 1% bovine serum albumin for the antibody diluent. Primary antibodies were used at a 1:500 dilution. Horseradish peroxidase-conjugated secondary antibodies (1:5,000 dilution, Transduction Laboratory) were used to visualize bound primary antibodies with the Supersignal chemiluminescence substrate (Pierce).
Time Course of Ad-cav-1 Expression in MTLn3 Cells-Infected cells grown in six-well plates were harvested at different times post-infection, and exogenous caveolin-1 levels were analyzed by immunoblotting. Alternatively, to monitor the time course of de novo caveolin-1 synthesis, metabolic labeling was performed on consecutive days post-infection. Briefly, cells were incubated in methionine-free ␣-MEM for 30 min and then labeled with [ 35 S]methionine (100 Ci/ml; NEN Life Science Products) for 2 h. After washing once with PBS, the cells were extracted for 1 h on ice with lysis buffer (see "Immunoblot Analysis"). COOHterminally Myc epitope-tagged caveolin-1 was immunoprecipitated from cell lysates using mAb 9E10, which is directed against the Myc epitope (11). Immunoprecipitates were then analyzed by SDS-PAGE and autoradiography.
Measurement of Anchorage-independent Growth-The growth in soft agar assay was performed as we described previously (23) with minor modifications. Infected MTLn3 cells grown in six-well plates were trypsinized and counted twice with a hemocytometer. Approximately 2.5 ϫ 10 4 cells were suspended in 3 ml of ␣-MEM containing 5% fetal bovine serum and 0.33% SeaPlaque low melting temperature agarose (FMC Bioproducts). The suspension was plated in a 60-mm dish containing a 2-ml layer of solidified ␣-MEM, 5% fetal bovine serum, and 0.5% SeaPlaque agarose. Three 60-mm dishes were used for each experimental condition. The cells were allowed to settle at the interface between these layers for 30 min at 37°C, and plates were hardened at room temperature for an additional 30 min before being returned to 37°C. Cells were fed every 5 days by overlaying with 2 ml of complete medium containing 0.33% SeaPlaque agarose. After 10 -15 days, the plates were examined under a microscope at low magnification (ϫ4 or ϫ6), and the colonies were counted. Experimental values represent the average number of colonies in the three 60-mm plates for each experimental condition; error bars represent the observed standard deviation among the three plates. When comparing uninfected MTLn3 cells with MTC cells, only foci with a diameter of Ͼ 400 M were considered; five independent fields were counted for each cell line.
Immunofluorescence Microscopy-Subconfluent MTLn3 cells grown on glass coverslips in 24-well plates were virally transduced 1 day before immunofluorescence staining. The staining procedure was performed as we described previously with minor modifications (11). Briefly, MTLn3 cells infected with Ad-tTA and either Ad-cav-1 or Ad-GFP were fixed for 30 min in PBS containing 2% para-formaldehyde, rinsed with PBS, and quenched with 50 mM NH 4 Cl for 10 min. The cells were then incubated in permeabilization buffer (PBS, 0.2% bovine serum albumin, 0.1% Triton X-100) for 10 min, washed with PBS, and labeled with a 1:400 dilution of anti-caveolin-1 rabbit IgG (pAb N-20; Santa Cruz Biotechnology) for 60 min. The GFP-infected cells were not labeled directly because they can be visualized by autofluorescence. After rinsing three times with PBS, a secondary antibody (7.5 g/ml) (lissamine-rhodamine-conjugated goat anti-rabbit IgG) was added for a period of 60 min. Cells were washed three times with PBS, and slides were mounted with Slow-Fade anti-fade reagent (Molecular Probes). A Bio-Rad MR600 confocal fluorescence microscope was used for visualization of the bound secondary antibody and GFP.
Lamellipod Extension Studies-Experiments were performed essentially as described previously by Segall and colleagues (32). MTLn3 cells were plated on collagen I-coated MATTEK tissue culture dishes at a density of 2,500 cells/cm 2 . One day later, the cells were virally infected and cultured for an additional 24 h. The infected cells were serum starved for 3 h in serum-free buffer (␣-MEM in 12 mM HEPES, pH 7.4, supplemented with 0.35% bovine serum albumin). Cells were then treated with 5 nM murine EGF (Sigma) in the same buffer for 3 min at 37°C. The cells were fixed immediately with 3.7% formaldehyde (in PBS) for 5 min at 37°C and permeabilized with 0.5% Triton X-100 (in PBS) for 20 min. After washing five times with PBS and incubating in blocking buffer for 30 min, the cells were stained with rhodaminephalloidin (1 M; to visualize F-actin) for 20 min in a humidified chamber. The cells were washed three times with blocking buffer and incubated with antibodies directed against the c-Myc epitope for 1 h to detect caveolin-1 expression. Unbound antibodies were removed by washing three times with blocking buffer followed by two washing with PBS. Bound antibodies were visualized with a fluorescein-conjugated secondary antibody probe. Coverslips were mounted on the MATTEK dishes with an anti-fade solution (Molecular Probes). Fluorescence images were acquired with an NA1.4 ϫ 60 objective using a Bio-Rad MRC-600 confocal microscope. Images of the distribution of F-actin and caveolin-1 were taken using rhodamine and fluorescein filters, respectively. Merged images from the same field were created using Adobe Photoshop.
Chemotaxis Assay-A 48-well microchemotaxis chamber (Neuroprobe) was used to study the chemotactic response to EGF, following the manufacturer's instructions. The chamber was washed thoroughly before use to remove any residue of chemoattractants. Nucleopore filters (8-m pore size) were coated with rat tail collagen I in PBS (27 g/ml) for 2 h. After filling the lower wells of the chamber with MEMH (␣-MEM, 12 mM HEPES, pH 7.4) containing 5 nM EGF, the filter membrane was laid carefully over the lower wells, and the whole chamber was assembled. An equal number of cells (2 ϫ 10 4 /well) were suspended in MEMH and loaded into the upper wells. The chamber was incubated for 3 h at 37°C. The cells that did not migrate across the filter were scraped, whereas the cells that traveled across were fixed in 3.7% formaldehyde (in PBS), washed in water, and stained in hematoxylin overnight. After a brief destaining in water, the filter was mounted between two glass plates using 90% glycerol (in PBS). The stained cells on the filter were counted under a microscope.
EGF Treatment of Adenovirus-infected MTLn3 Cells-MTLn3 cells were infected with Ad-cav-1 or Ad-GFP, with or without Ad-tTA. One day post-infection, cells were incubated with 5 nM murine EGF for 30 min. The cells were lysed in hot 1% SDS lysis buffer with 50 mM Tris-HCl, pH 7.4, and the cell lysates were analyzed by immunoblot-ting. Activation of ERK was detected with a phospho-specific antibody probe that only recognizes the active phosphorylated form of ERK; total expression of ERK was monitored with a phospho-independent antibody probe (New England Biolabs). Untreated cells (without EGF treatment) were also used to determine basal levels of phosphorylated and total ERK.

Reduction of Caveolin-1 and -2 Levels in a Metastatic Mammary Adenocarcinoma Cell Line (MTLn3 Cells) Cells-Caveo-
lin-1 levels have been shown to be reduced or absent in mammary tumors and tumor-derived cell lines (18,22,39). The MTLn3 and MTC cell lines are, respectively, motile and nonmotile sublines derived from a single rat mammary adenocarcinoma cell line (13762NF; first described in Refs. 25 and 36). As shown in Fig. 1 (top panel), caveolin-1 is expressed in the non-motile MTC subline, whereas it is clearly absent in motile MTLn3 cells. Also, note that the widely used human mammary carcinoma-derived cell lines, MCF-7 and T-47D, show an absence of caveolin-1 expression compared with normal human mammary epithelial cells and NIH 3T3 cells (a cell line with robust caveolin-1 expression). In addition, the use of a highly sensitive anti-caveolin-1 rabbit IgG (pAb N-20) also failed to detect any caveolin-1 protein expression in the same cell lines (Fig. 1, middle panel). Although reductions in caveolin-2 levels do not necessarily correlate with a transformed phenotype (23,40), we observed down-regulation of caveolin-2 in all of the mammary carcinoma-derived cell lines, with the exception of non-motile MTC cells (Fig. 1, bottom panel).
Characterization of a Recombinant Caveolin-1 Adenoviral Vector as a Modality for Efficient Gene Transfer of the Caveolin-1 cDNA to MTLn3 Cells-The down-regulation of caveolin-1 in MTLn3 cells led us to study the effect of caveolin-1 overexpression in these cells via an adenovirus-based approach. The vector used is shown schematically in Fig. 2A. The COOHterminally Myc-tagged canine caveolin-1 cDNA was placed under the control of a CMV immediate early minimal promoter preceded by a heptamer of tetO sequences. Protein expression could then be induced by co-infection of the caveolin-1 adenovirus with Ad-tTA. The tTA, a fusion between the bacterial tet repressor and the activation domain of the herpes simplex virus protein VP16, transcriptionally activates promoters containing tetO elements in the absence of tet (41,42). Fig. 2B shows robust expression of caveolin-1 in MTLn3 cells co-infected with Ad-cav-1 and Ad-tTA. Note the presence of a dose response for caveolin-1 expression when using varying m.o.i. values (lanes 1-3) and that in the absence of Ad-tTA, caveolin-1 expression is prevented (lane 4). As controls in all of our experiments, we utilized a GFP adenovirus (Ad-GFP) engineered in the same manner as for caveolin-1. Note that the expression pattern for GFP qualitatively mimics that of caveolin-1 (lanes 6 -9). To assess the expression of the caveolin-1 and control GFP proteins in vivo, we performed immunofluorescence microscopy on infected MTLn3 cells (Fig. 2C). As expected, caveolin-1 displayed a membrane-localized punctate staining pattern that is characteristic of caveolae membranes (7,9,10,43,44), whereas GFP showed a homogenous cytoplasmic and nuclear staining pattern.
Given the versatility of this inducible adenoviral vector system, we were interested in the kinetics of exogenous caveolin-1 expression in MTLn3 cells. Immunoblots of cells co-infected with either the caveolin-1 or GFP adenoviruses and the tTA adenovirus at varying time points indicated maximal protein levels at ϳ 1-2 days post-infection, with high levels of expression well beyond the 50-h time point (Fig. 3A). Pulse labeling of MTLn3 cells with [ 35 S]methionine and immunoprecipitation of the Myc-tagged caveolin-1 protein allowed us to determine the duration of de novo caveolin-1 production post-infection. Fig.  3B demonstrates that although most of the new protein is synthesized in days 1-3, caveolin-1 production can still be detected into the 7th day post-infection. Because we have determined the half-life of caveolin-1 to be on the order of 24 -36 h, 2 such adenoviral gene transfers could theoretically maintain high levels of caveolin-1 expression for 1 week or longer.
Adenoviral Delivery of the Caveolin-1 cDNA Blocks Anchorage-independent Growth in MTLn3 Cells-Previous studies have shown that there is an inverse correlation between caveolin-1 expression and the ability of transformed cell lines to form colonies in soft agar (23,40). These studies were performed using either caveolin-1-inducible or constitutively expressing NIH 3T3 cell lines. To extend these findings to MTLn3 cells, we performed soft agar assays using the adenoviral gene delivery  6 -9). Note that as the m.o.i. values are increased, there is a dose-dependent increase in expression. Importantly, in the absence of the co-requisite Ad-tTA, neither caveolin-1 nor GFP expression is induced (lanes 4 and 9, respectively). Each lane contains an equal amount of total protein. C, immunofluorescence of MTLn3 cells expressing either caveolin-1 or GFP. Upper panels, cells co-infected with Ad-cav-1/Ad-tTA were fixed, permeabilized, and incubated sequentially with the anti-Myc IgG and a rhodamine-labeled anti-rabbit secondary antibody. The cells were viewed by employing a confocal microscope with settings for both phasecontrast and fluorescence. Lower panels, cells co-infected with Ad-GFP/ Ad-tTA were processed similarly for microscopy with the exception of antibody treatment. Note that caveolin-1 assumes a characteristic punctate membrane distribution, whereas GFP is present homogeneously in both the cytosol and the nucleus. system to maintain prolonged caveolin-1 expression.
First, we compared the ability of uninfected MTLn3 cells (caveolin-1-negative) and MTC cells (caveolin-1-positive) to undergo anchorage-independent growth. Fig. 4 shows that MTLn3 cells form foci ϳ5-fold more efficiently than MTC cells; in addition, the colonies formed by uninfected MTLn3 cells are much larger than MTC colonies.
Next, we evaluated the ability of caveolin-1-transduced MTLn3 cells to undergo anchorage-independent growth. Fig.  5A shows the number of colonies for MTLn3 cells infected with varying m.o.i. values of Ad-cav-1 or Ad-GFP, with or without Ad-tTA. The number of colonies presented is normalized to the mean value from mock-infected cells. Note that caveolin-1 expression produced a dose-dependent reduction (ϳ2-3 fold) in the number of colonies formed as compared with the GFP control. Furthermore, in the absence of the co-requisite Ad-tTA, the Ad-cav-1-infected cells formed colonies as readily as the Ad-GFP control or mock-infected cells. Fig. 5B displays representative MTLn3 colonies observed in soft agar. Caveolin-1-expressing cells not only show a reduced number of colonies, but also the size of these colonies is reduced dramatically.

Restoration of Caveolin-1 Expression in MTLn3 Cells Effectively Blocks EGF-stimulated Lamellipod Extension and Cell
Migration-Segall and colleagues have used MTLn3 cells extensively, in combination with the chemoattractant EGF, to study the mechanics of lamellipod extension and cell migration (32). Cellular flattening and lamellipod growth are readily seen in EGF-stimulated MTLn3 cells (45), a process concomitant with actin polymerization and localized filament formation (32,33). Thus, we next evaluated the effects of caveolin-1 expression on the motile phenotype of MTLn3 cells, as measured via EGF-stimulated lamellipod extension and cell migration. Fig. 6A shows EGF-treated MTLn3 cells transduced with the caveolin-1-containing adenoviral vector (plus or minus the corequisite Ad-tTA adenovirus), the GFP control, and mock-infected cells. All cells were co-stained with rhodamine-phalloidin to visualize F-actin fibers directly. Note the formation of F-actin fibers and anisotropic lamellipod extension in the mock-infected, Ad-cav-1 alone, and GFP control cells. In striking contrast, cells recombinantly expressing the caveolin-1 protein (Ad-cav-1 plus Ad-tTA) are smaller and show reduced or absent lamellipod extension.
This difference in morphology between caveolin-1-expressing and non-expressing cells can be seen more readily in Fig. 6B  (upper panel), showing a color overlay of the F-actin and caveolin-1 staining from Fig. 5A. Note that that the similar overlay for the GFP-expressing cells does not show compromised Factin polymerization or loss of lamellipod extension (Fig. 6B,  lower panel).
For the study of chemotactic responses and their effects on cell migration, we used a well characterized microchemotaxis chamber system (32). In this assay, MTLn3 were cells were placed over a porous filter and presented with the EGF stimulus in an adjacent compartment. After 3 h, cells that migrated to the EGF compartment were counted. Fig. 7 shows the quantitation of this migration event for cells transduced with caveolin-1 and GFP adenoviruses, plus and minus co-infection with the tTA adenovirus required for expression. All data are normalized to the number of cells that migrated in the mockinfected group (percent control). Note that MTLn3 cells recombinantly expressing the caveolin-1 protein (Ad-cav-1 plus Ad-tTA) show a dose-dependent reduction (ϳ3-4-fold) in cell migration compared with mock-infected, Ad-cav-1 alone, and GFP control cells.
It is now well established that EGF potently activates the p42/44 MAP kinase cascade, along with other signaling pathways. Because several independent lines of evidence now support the idea that caveolin-1 functions as a negative regulator of the p42/44 MAP kinase cascade (19,24), we investigated the potential negative regulatory role of caveolin-1 in EGF-mediated activation of the p42/44 MAP kinase cascade in MTLn3 cells. MTLn3 cells were transduced with caveolin-1 and GFP adenoviral vectors at varying m.o.i. values, and activation of the p42/44 MAP kinase pathway was assessed by immunoblotting employing phospho-specific antibodies that recognize only the activated form of ERK. Fig. 8A shows that caveolin-1 expression, but not the GFP control, is able to reduce EGF-induced activation of ERK significantly by ϳ2-3-fold (upper panel, lanes [3][4][5][6][7][8]. Note that in the absence of EGF, ERK is not activated at base line (upper panel, lanes 1 and 2) and that the total amount of ERK protein is equivalent in all infected and control cells, as visualized with a phospho-independent antibody (Fig. 8A, lower panel). Importantly, note that the level of EGF-R expression is not altered by transduction with the caveolin-1 or GFP adenoviral vectors (Fig. 8B).

DISCUSSION
Caveolin-1 is a potent negative regulator of a variety of mitogenic signaling pathways. The loss or reduction of caveolin-1 expression appears to be a frequent event in transformed cells and during tumorigenesis. Recombinant expression of caveolin-1 in oncogenically transformed NIH 3T3 cells and a human mammary carcinoma-derived cell line (T-47D) abrogates their anchorage-independent growth phenotype (39,40). Anchorage-independent cell growth is a common phenomenon in malignant transformation and is related to a loss of contact inhibition, lamellipod extension, and cell migration/invasiveness. Therefore, it is surprising that the possible involvement of caveolin-1 in negatively regulating lamellipod formation and cell migration has not yet been addressed.
Here, we have shown that MTLn3 cells (a motile rat mammary adenocarcinoma-derived cell line) lack caveolin-1 expression, in contrast to MTC cells (the non-motile counterpart of MTLn3 cells), which continue to express significant levels of caveolin-1. To restore the expression of caveolin-1 in MTLn3 cells, we employed a tet-based adenoviral vector system to deliver the caveolin-1 cDNA and induce caveolin-1 protein expression over prolonged periods. Using this modality, we show that recombinant expression of caveolin-1 in MTLn3 cells ab-rogates their anchorage-independent growth, reducing both the number and overall size of the colonies formed in soft agar. Using EGF as a chemoattractant, we showed that caveolin-1 expression effectively blocks the motility of MTLn3 cells. More specifically, MTLn3 cells expressing caveolin-1 exhibited a significant reduction in the EGF-induced formation of F-actin stress fibers at lamellipodia. In addition, caveolin-1 expression dramatically inhibited the migration of MTLn3 cells toward an EGF source using a microchemotaxis chamber. Finally, as we show that caveolin-1 expression blocks EGF-stimulated activation of the p42/44 MAP kinase cascade in MTLn3 cells, inhibition of this pathway by caveolin-1 may partially explain the ability of caveolin-1 to negatively regulate cell motility.
The inhibitory effects of caveolin-1 on MTLn3 lamellipod formation and cellular motility are indeed intriguing. Metastasis of primary tumors is a complex process that is related to a number of factors, including extracellular matrix attachment, matrix proteolysis and basement membrane disruption, migration toward a blood vessel source (a process concomitant with angiogenesis), and tumor cell dissemination and seeding of distant sites (for review, see Ref. 46). Interestingly, there are other indications that caveolin-1 may play an important role in this process. Vascular endothelial growth factor, a well known activator of angiogenesis, causes a reduction in caveolin-1 levels in endothelial cells. This process is readily blocked by an- Evaluation of cellular migration was performed using a microchemotaxis chamber system. The number of cells that crossed the membrane in 3 h (i.e. underwent chemotaxis toward EGF) was normalized to that of mock-infected cells. The values presented are averages for three to six wells from two separate experiments. Note that the caveolin-1expressing cells (Ad-cav-1/Ad-tTA) have a significantly reduced chemotactic response. This inhibitory effect was accentuated at higher doses of Ad-cav-1.

FIG. 8. Caveolin-1 inhibits EGF-mediated stimulation of the p42/44 MAP kinase pathway but not expression of EGF-R itself.
A, p42/44 MAP kinase activation. MTLn3 cells were transduced with Ad-cav-1 or Ad-GFP, in the presence or absence of the transactivator, Ad-tTA. One day later, MTLn3 cells were treated with either EGF (5 nM; lanes [3][4][5][6][7][8] or buffer alone (lanes 1 and 2) for 30 min. Cells were then lysed in hot sample buffer, and samples were subjected to immunoblot analysis with anti-phospho-ERK IgG to visualize activated ERK (upper panel) or anti-ERK IgG to visualize the total amount of ERK expression (lower panel). Note that EGF-mediated activation of ERK is inhibited by ϳ2-3-fold in caveolin-1-expressing cells (lanes 3 and 4), but not in control cells expressing GFP (lanes 6 and 7). giogenic antagonists (47). In a series of studies, Giancotti and colleagues (48,49) implicate caveolin-1 as an integrin-associated protein that is important for regulating integrin signaling and anchorage-dependent cell growth. Alterations in integrin signaling often lead to a loss of cell adhesion (a prerequisite for anchorage-independent proliferation and migration in tumor cells) (for review, see Ref. 50).
Are the effects of caveolin-1 on lamellipod extension and chemotaxis caused by its involvement in integrin signaling or because of its inhibitory interaction with the EGF-R and downstream pathways? Segall and colleagues (32)(33)(34) have studied the effects of EGF on lamellipod extension and chemotaxis, showing that EGF-induced actin polymerization at the leading edge of lamellipodia plays a significant role in their extension and cellular motility. Although the relevant effectors and crosstalk between EGF-R and actin polymerization are unknown, caveolin-1 has been shown to affect EGF-R signaling at numerous steps. Caveolin-1 can interact directly with EGF-R and functionally inhibit its activity in vitro and in vivo (17,19). Caveolin-1 also has effects on some of the primary downstream effectors of EGF-R and is able to interact directly with H-Ras (11,12) and to inhibit the activation of members of the p42/44 MAP kinase cascade (19). In addition, reports from several other laboratories have demonstrated that activation of the p42/44 MAP kinase cascade is required for cell migration (51)(52)(53)(54)(55).
On the other hand, integrins act to connect the extracellular matrix with the cytoskeleton. An accumulation of actin filaments, actin-associated proteins (e.g. ␣-actinin, talin, vinculin), various signaling molecules (e.g. focal adhesion kinase and Src-family kinases), and integrins form what are known as focal adhesions (56). Interestingly, caveolin-1 is known to interact functionally with components of this complex as well. Caveolin-1 was first identified as a transformation-dependent v-Src substrate in Rous sarcoma virus-transformed cells (57). Both c-Src and other Src family tyrosine kinases co-purify with caveolae (43,58) and interact functionally with caveolin-1 (11) by phosphorylating its Tyr-14 residue in vitro and in vivo (59,60). In addition, caveolin-1 may function as a linker to couple the integrin ␣-subunit with its effectors Fyn and Shc, an interaction that is critical for anchorage-dependent growth (49).
In support of these observations, cellular depletion of caveolin-1 via an antisense approach leads to disruption of integrin signaling (61). The ligand activation of integrin receptors is often the initiating response in cytoskeletal rearrangements, alterations of which are pivotal for tumor cell motility and invasion (62). Although the placement of caveolin-1 in integrin signaling is still rudimentary, caveolin-1 could possibly interact with any one of the components of focal adhesions to affect the formation of actin stress fibers and by extension, cellular motility. Further studies will be necessary to provide a complete mechanistic understanding of exactly how caveolin-1 regulates anchorage-independent growth/contact inhibition and cell motility.