Phosphorylated Caveolin-1 Regulates Rho/ROCK-Dependent Focal Adhesion Dynamics and Tumor Cell Migration and Invasion

Rho/ROCK signaling and caveolin-1 (Cav1) are implicated in tumor cell migration and metastasis; however, the underlying molecular mechanisms remain poorly defined. Cav1 was found here to be an independent predictor of decreased survival in breast and rectal cancer and significantly associated with the presence of distant metastasis for colon cancer patients. Rho/ROCK signaling promotes tumor cell migration by regulating focal adhesion (FA) dynamics through tyrosine (Y14) phosphorylation of Cav1. Phosphorylated Cav1 is localized to protrusive domains of tumor cells and Cav1 tyrosine phosphorylation is dependent on Src kinase and Rho/ROCK signaling. Increased levels of phosphorylated Cav1 were associated with elevated GTP-RhoA levels in metastatic tumor cells of various tissue origins. Stable expression and knockdown studies of Cav1 in tumor cells showed that phosphorylated Cav1 expression stimulates Rho activation, stabilizes FAK association with FAs, and promotes cell migration and invasion in a ROCK-dependent and Src-dependent manner. Tyrosine-phosphorylated Cav1, therefore, functions as an effector of Rho/ROCK signaling in the regulation of FA turnover and, thereby, tumor cell migration and invasion. These studies define a feedback loop between Rho/ROCK, Src, and phosphorylated Cav1 in tumor cell protrusions, identifying a novel function for Cav1 in tumor metastasis that may contribute to the poor prognosis of some Cav1-expressing tumors. [Cancer Res 2008;68(20):8210–20]

The small GTPases Rho, Rac, and Cdc42 control signaling pathways linking cell surface receptors to actin cytoskeleton dynamics and associated focal adhesion (FA) assembly and disassembly. Upon activation, in response to extracellular signals, Rac and Cdc42 stimulate actin polymerization and nascent FAs at the cell periphery resulting in the formation of actin-based pseudopodial protrusions. Rho regulates FA maturation and stabilization via linkage to actin stress fibers and the actomyosin contractile apparatus (1). However, contrasting its role in FA stabilization, Rho, and particularly Rho signaling via Rho kinase (ROCK), is closely associated with tumor cell migration and metastasis.

RhoC has been shown to be critically associated with tumor metastasis (2). ROCK expression in colon carcinoma is associated with increased tumor cell dissemination, whereas ROCK inhibition reduces the migration and invasion of malignant astrocytoma cells (3, 4). Activated Rho is localized to the leading edge of motile cells, including metastatic breast carcinoma cells (5, 6).⁷

Rho/ROCK expression is associated with alterations in the properties and signaling of membrane protrusions of motile cells (5, 7–9). A Rho/ROCK-dependent rounded, blebbed, nonproteolytic mode of tumor cell invasion has been described to promote invasion via matrix deformation and modulation of cell-matrix adhesions (10, 11). However, effectors of Rho/ROCK function in tumor cell migration have yet to be identified.

Caveolin-1 (Cav1) is a multifunctional membrane protein responsible for caveolae formation that is also involved in the regulation of signal transduction and raft-dependent endocytosis (12). Whereas Cav1 is a negative regulator of cytokine receptor signaling, various studies report of contrasting tumor suppressor and promoter functions depending on the tumor cell model or the tumor type studied (13, 14). Indeed, contrasting a tumor suppressor function of Cav1, its expression has long been associated with a poor prognosis and reduced survival for prostate cancer patients (15). Furthermore, tissue microarray analysis (TMA) of invasive breast carcinomas in this study is consistent with reports identifying Cav1 overexpression to be associated with a basal-like phenotype and predicting a worse breast cancer patient prognosis (16, 17). TMA analysis of rectal cancer showed that Cav1 expression was also associated with poor survival. In colon cancer, patients did not reveal an association of Cav1 with survival, but did identify an association between Cav1 overexpression and the presence of distant metastatic disease. These observations highlight the varied implications of Cav1 expression on tumor progression in various cancer cell models and are indicative of a role for Cav1 in tumor metastasis. However, the function of Cav1 in tumor cell migration and metastasis remains poorly defined.

Cav1 was first identified as the major tyrosine-phosphorylated protein in Rous sarcoma virus transformed chicken embryo fibroblasts (18). Tyrosine phosphorylated Cav1 (pY14Cav1) has recently been identified as an activator of Rho and regulator of directional migration of mouse embryo fibroblasts (19). Cav1 interacts with integrins (20, 21) and has been shown to regulate FA domain organization, dynamics, and turnover (22, 23). We show here that expression of pY14Cav1 is critically associated with Rho/ROCK and Src-dependent regulation of tumor cell motility and invasion. Using well-characterized metastatic human tumor cell lines, pY14Cav1 expression is localized to membrane protrusions of tumor cells and associated with Rho activation, FAK stabilization in FAs, and enhanced tumor cell migration and invasion. pY14Cav1 function as an effector of Rho/ROCK signaling to promote late-stage tumor progression and metastasis may thereby explain its association with a poor prognosis in various human tumors.

Antibodies and reagents. Rabbit anti-Cav1/2 and mouse anti-pY14Cav1 antibodies were purchased from Transduction Laboratories, rabbit anti-c-Src and anti-RhoA antibodies from Santa Cruz, mouse mitochondrial-Hsp70 (mHSP70) and p-Src (Y418) antibodies from Ab-Biochem, and anti-Myc tag mouse antibody and RhoA-GTP assay kit from Upstate Biotechnology. Alexa-488–conjugated and Alexa-647–conjugated secondary antibodies were purchased from Molecular Probes. Mouse anti–β-actin, rabbit anti-ROCK1, PP2 and Y27632 were from Sigma.

Plasmids and small interfering RNAs. Cav1-mRFP-wt, Cav1Y14F-mRFP, Cav1Y14D-mRFP, and FAK-GFP have been described previously (23). Wild-type, dominant-negative (DN) and dominant active (DA) RhoA and Rac1 plasmids were a generous gift from Nathalie Lamarche (McGill University). Validated human c-Src (SMARTpool reagent M-003175-01-0005) and ROCK1 small interfering RNAs (siRNA; SMARTpool reagent M-003536-01-0005), as well as control siRNAs and siGLO Red (D-001630-02-05), were purchased from Dharmacon.

Cell lines. Human MDA-231, MCF-7, MCF-10A, MDA-435, HT29, HCT116, SW480, LS174T, DU145, PC3, HEPG2, LnCAP, and mouse CT26 cell lines were from American Type Culture Collection and maintained in complete RPMI 1640 supplemented with 10% fetal bovine serum (FBS). MDA-231 Cav1–short hairpin RNA (shRNA) stable cell line (24) was maintained in medium supplemented with blasticidine (Invitrogen). Murine sarcoma virus (MSV)–Madin-Darby canine kidney cells (MDCK)–INV cells were grown in DMEM supplemented with 10% FBS, as previously described (8). Stable MDA-435 cell lines expressing dsRed, Cav1-mRFP, or Cav1(Y14F)-mRFP were prepared by cotransfecting dsRed, Cav1-mRFP, or Cav1Y14F-mRFP expression vectors with pcDNA3 as a selection marker using Effectene (Invitrogen). Neomycin-resistant cells were selected for 15 d against 400 μg/mL of geneticin (Life Technologies, Invitrogen), and resistant colonies trypsinized and sorted for mRFP positives by fluorescence-activated cell sorting. Pooled mRFP-positive cells were allowed to recover and expanded in complete medium supplemented with geneticin. All cell lines were passaged at least twice after recovery from frozen stocks before initiating experiments and maintained in culture for a maximum of 8 to 10 passages to minimize phenotypic drift. Cells were plated at 80% confluency and allowed to grow for 24 h before treatment with 10 μmol/L PP2 or 20 μmol/L Y27632 for the indicated times. Transient transfections with Effectene (Qiagen) or X-tremeGENE (Roche) were performed using standard protocols and cells used after 24 or 48 h, as indicated. Two micrograms of control or targeted siRNAs were transfected with X-tremeGENE reagent (Roche) following the manufacturer's protocol; cells were allowed to grow for 48 h before assessing gene knockdown by Western blotting.

Pseudopodia preparation and Western blotting. Pseudopodial preparation, isolation, and protein extraction were as previously described (8). For Western blotting, cells were allowed to grow to 80% confluency and washed in cold PBS, lysates were prepared, and Western blots were performed as previously described (24). RhoA-GTP pull-downs were performed according to the supplier's protocol (Upstate Biotechnology).

Immunofluorescence labeling. Cells were fixed with 3% paraformaldehyde for 15 min at room temperature, rinsed with PBS, permeabilized with 0.1% Triton X-100 in PBS/CM containing 0.5% bovine serum albumin (BSA), and then incubated with primary and fluorescent secondary antibodies in PBS/CM containing 0.2% BSA. After labeling, the coverslips were mounted in Airvol (Air Products, Inc.), and images were collected with the 60× or 100× planapochromat objectives (NA 1.35) of an FV1000 Olympus confocal microscope.

FA dynamics and live cell imaging. Cells plated at low density in eight-well IDIBI chambers were cotransfected with Cav1-mRFP and β-actin-GFP. After 24 h, images of cotransfected cells were acquired every 30 s for 30 min at 37°C with the 60× planapochromat objective of an FV1000 Olympus confocal microscope. Cells plated at low density in an eight-well IDIBI chamber were transfected with FAK-GFP for 24 h. Fluorescence recovery after photobleaching (FRAP) was performed at 37°C with the 60× planapochromat objective of a FV1000 Olympus confocal microscope (23). Two prebleach events were acquired followed by a single bleach event with the 405 line of the SIM scanner. Fluorescence recovery was followed at 4-s time intervals until intensity reached a plateau and normalized to the prebleach intensity. Relative recovery rates for FAK-GFP at FAs were compared using the half-time for recovery of fluorescence toward the asymptote. Intensity in the bleached area before bleaching and after recovery was used to calculate mobile and immobile fractions. Graphs are representative of a minimum of three independent experiments, in which between 6 and 15 FAs were bleached.

Migration, invasion, adhesion, colony formation, and cell proliferation. Cells were trypsinized, counted, and transferred to uncoated (migration) or Matrigel-coated (invasion) 8-μm cell culture inserts (BD Falcon) in medium containing 2% serum and the assembly placed into 24-well plates containing complete medium. After 16 h, nonmigrating and noninvasive cells were removed from the top of the filter with a cotton swab, and migrating cells on the bottom of the filter were fixed with methanol acetone, stained with 5% crystal violet, and counted. Colony formation assay was performed by plating 3,000 cells per line per well and selecting for 2 wk, followed by fixing, staining with 5% crystal violet, and scoring colonies with >20 cells. Cell proliferation assay was performed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (Sigma). CytoMatrix cell adhesion strips (Chemicon International, Inc.) coated with human collagen type I, human collagen type IV, human fibronectin, human laminin, and human vitronectin were used according to the manufacturer's recommendations.

Human tumor TMAs. Four hundred thirty-eight sequential archival cases of invasive breast carcinoma, 127 cases of colon carcinoma, and 131 cases of rectal carcinoma with available paraffin blocks were identified for TMA construction. Clinical data on all patients was retrospectively collected from patient hospital charts. The study was approved by the Clinical Research Ethics Boards of the University of British Columbia and Providence Health Care. Anti-Cav1 (Santa Cruz) dilutions were validated using cores prepared from cell pellets to distinguish the higher Cav1 expression, as determined by Western blot, of MDA-231 cells from MDA-435, MCF7, and MCF10A cells (24). Sections were stained with a 1:1,000 dilution in EDTA buffer (pH 8.0). Cav1 was considered overexpressed if ≥25% of invasive carcinoma cells exhibited membranous staining. Staining and scoring for ER, PR, and HER-2 were as previously described (24). TMA construction was carried out as previously described (25), and all samples were scored by two pathologists blinded to all patient clinical data.

Breast cancer TMA. Median patient follow-up was 15 y, and all patients had newly diagnosed stages I to III invasive breast cancer. Clinicopathologic data collected included patient age, sex, follow-up and survival, lymph node status (negative versus positive), tumor size, grade, histology, ER status, PR status, and HER-2 status. All data were logged onto a standardized score sheet matching each TMA section (Microsoft Excel), processed using TMA-Deconvoluter 1.07 software (26), and analyzed by SPSS 11.0 statistical software. Correlation analysis used the bivariate two-tailed Spearman nonparametric correlation test. Differences were considered significant when P < 0.05. The Kaplan-Meier method was used for survival analysis with log-rank test to determine significance in survival differences. For the multivariate analysis, we used a proportional hazards model (Cox regression model).

Colon and rectal cancer TMAs. Mean patient follow-up was 3.6 y for the colon TMA and 4.9 y for the rectal TMA, and all patients had newly diagnosed stages I to IV colon cancer. Clinicopathologic data collected included patient age, sex, follow-up and survival, urgency of surgery, tumor location, size, differentiation, and margin status, presence of vascular invasion or lymphatic invasion, and lymph node status (negative versus positive). All data was logged onto a standardized score sheet matching each TMA section (Microsoft Excel) and deconvoluted using custom software (developed by O.G. with Perl programming language). Deconvoluted TMA scoring data was transferred to a master study database that contained all clinical and pathologic data. Clinicopathologic variables and receptor overexpression were examined for their prognostic significance by using univariate and multivariate survival analyses using the log-rank test and stepwise logistic regression analysis, respectively. Survival outcomes examined included overall survival, disease-specific survival, and disease-free survival. Disease-free survival included stage IV patients if they underwent successful resection of all metastatic diseases. Association between receptor expression and prognostic variables was determined by contingency table statistics (for categorical variables) and the Mann-Whitney U test (for continuous variables). Correlational analysis was determined by Spearman correlation. All tests were two-tailed and considered significant at P < 0.05. P values were corrected for multiple testing by Benjamin and Hochberg method where appropriate. All statistics were performed using the SPSS 13.0 statistical software.

TMA analysis of Cav1 expression in breast, colon, and rectal carcinoma. Cav1 overexpression in breast cancer was identified as a highly significant, independent predictor of decreased survival (Fig. 1A). Twenty-six of 343 interpretable cases presenting Cav1 labeling showed a highly significant association with decreased disease-specific survival and nodal involvement (P = 0.007). In the Cox model with such variables as Cav1, ER, HER2, tumor size, Nottingham grade, and age of patients at the moment of diagnosis, Cav1 was an independent prognostic factor (hazard ratio = 3.766, P = 0.00036).

For colon cancer patients, Cav1 overexpression did not predict patient outcome (Fig. 1B) highlighting the varied role of Cav1 expression in patient survival in different tumor types (13, 14). However, we did observe a significant association (P = 0.045) between Cav1 expression in colon cancer and the presence of distant metastatic disease at presentation (Fig. 1B , inset), supporting an association between Cav1 expression and cancer progression and metastasis.

For rectal cancer patients, Cav1 overexpression was associated with decreased years of disease-free survival (P = 0.005). Furthermore, we observed a significant association between Cav1 expression in rectal cancer and the development of distant metastasis (disease recurrence, P = 0.049; Fig. 1C).

pY14Cav1 is expressed in select metastatic cell lines. Caveolin (Cav1/2) expression was profiled in MCF-10A, MCF-7, MDA-231, MDA-435, HEPG2, CT26 (mouse), LnCAP, DU145, PC3, SW480, LS174T, HCT116, and HT29 cell lines. As shown in Fig. 2A, Cav1/2 expression was significantly increased in metastatic MDA-231, PC3, DU145, CT26, and HCT116 cells relative to the other cell lines. Interestingly, pY14Cav1 levels were selectively detected in breast MDA-231, colon HCT116, and prostate PC3 cells, where it was associated with either elevated Cav1 levels and/or elevated expression of phosphorylated Src (pSrc; Supplementary Fig. S1A). In the previously characterized MSV-MDCK-INV cell line that exhibits well-defined protrusive pseudopodial domains (8), pY14Cav1 levels were increased relative to the polarized MDCK epithelial cell line (Fig. 2A).

Using a protocol to purify pseudopodial domains projecting through 1 μm filter pores (8), we determined whether Cav1/2 and pY14Cav1 were localized to tumor cell protrusions. As seen in Fig. 2B for MSV-MDCK-INV and MDA-231 cells, both pY14Cav1 and pSrc are significantly enriched in the pseudopodial fraction relative to the cell body. Importantly, mitochondrial HSP70 is predominantly expressed in the cell body serving as a marker of the purity of the pseudopodial fraction. Cav1/2 is also found in the pseudopodial fraction localizing caveolin to tumor cell protrusions.

By immunofluorescence, Cav1/2 distribution in MSV-MDCK-INV, HCT116, PC3, and MDA-231 cells did not exhibit a polarized distribution and extended to peripheral cellular regions (Supplementary Fig. S1B). To verify the distribution of Cav1 to protrusive cellular domains, we cotransfected the MDA-435 cell line that expresses reduced levels of endogenous Cav1 (Fig. 2A) and few caveolae (24), as well as DU145 cells that express Cav1 but not pY14Cav1 (Fig. 2A), with Cav1-mRFP or Cav1Y14F-mRFP and β-actin-GFP. Time lapse imaging shows clearly that Cav1-mRFP is present in protrusive lamellipodial domains defined by β-actin-GFP expression (Fig. 2C, and D, left; Supplementary Videos 1 and 3). In contrast, mutant Cav1Y14F-mRFP is excluded from the β-actin-GFP–rich leading edge of moving cells (Fig. 2C  and D, right; Supplementary Videos 2 and 4). Kymograph analysis showed that wild-type (WT) Cav1, but not Cav1Y14F, was expressed throughout the lamellipodial region. These data suggest that pY14Cav1 is targeted to tumor cell protrusions.

Cav1 tyrosine phosphorylation is Rho/ROCK-dependent and Src-dependent. pY14Cav1 expression has been shown to be associated with Rho activation in mouse embryo fibroblasts (19). As shown in Fig. 3A, significantly increased GTP-RhoA levels were detected in breast MDA-231, prostate PC3, and colon HCT116 relative to MDA-435, DU145, and HT29 cells, corresponding to the elevated expression of pY14Cav1 in those cells (Fig. 2A). Total RhoA levels were similar in all the cell lines. We then tested the ability of DA and DN forms of RhoA and Rac1 to affect pY14Cav1 expression in MDA-231 and PC3 cells. Both DN-RhoA and DA-Rac1 selectively reduced pY14Cav1 levels of transfected MDA-231 and PC3 cells by Western blot analysis (Fig. 3B). Cav1 tyrosine phosphorylation is therefore downstream of activated Rho and regulated by differential activation of the Rho/Rac GTPases.

Rho/ROCK activation is associated with pseudopod protrusion in MSV-MDCK-INV cells (8, 27), and Cav1 is a known substrate of Src kinase (18). We therefore assessed the role of ROCK and Src signaling on Cav1 phosphorylation. As seen in Fig. 3C, treatment with 10 μmol/L PP2, a Src family kinase inhibitor, or 20 μmol/L Y27632, a ROCK inhibitor, significantly reduced Cav1 phosphorylation by Western blot in MSV-MDCK-INV, MDA-231, PC3, and HCT116 cells. Transfection of MDA-231 and PC3 cells with targeted siRNA selectively reduced expression levels of c-Src and ROCK1, whereas control siRNA had no effect on the expression level of the other targeted protein or on the expression of Cav1/2 or β-actin (Fig. 3D). Both c-Src and ROCK1 siRNA significantly reduced pY14Cav1 compared with control siRNA (Fig. 3D), confirming the data obtained with specific inhibitors (Fig. 3C). ROCK1 siRNA mediated knockdown also inhibited Src phosphorylation, suggesting that ROCK1 may regulate Src activity in these cells (Fig. 3D). This suggests that c-Src–mediated phosphorylation of Cav1 is linked to Rho/ROCK signaling.

Increased FA dynamics in pY14Cav1-expressing tumor cell lines is Src-dependent and ROCK-dependent. Rho/ROCK signaling has been associated with increased tumor cell invasivity via a nonproteolytic, amoeboid mode of tumor cell migration (11). However, zymogram studies did not show a correlation between matrix metalloproteinase (MMP) expression and pY14Cav1 expression or Rho activation state in these tumor cells (data not shown). Indeed, MDA-231 cells have recently been shown to follow a proteolytic, mesenchymal migratory mode (28) and HCT116 cells to exhibit ROCKII-dependent regulation of invasion, MMP-2 and MMP-13 activity, and invadopodia formation (29). Aside from increased binding of prostate PC3 cells to collagen types I and IV relative to DU145 cells, we detected no significant changes in adhesion to extracellular matrix components between the tumor cell lines studied (Supplementary Fig. S1C). However, using a Boyden chamber assay, we detected increased migration of pY14Cav1 expressing MDA-231, PC3, and HCT116 cells relative to MDA-435, DU145, and HT29 cells that do not express elevated pY14Cav1 levels (Fig. 4A). The migration of pY14Cav1-expressing MDA-231, PC3, and HCT116 cells, but not of MDA-435, DU145, and HT29 tumor cells, was sensitive to ROCK and Src inhibition (Fig. 4A).

We previously observed that pY14Cav1 stabilizes FAK exchange in FAs leading to enhanced FA turnover (23). The presence of pY14Cav1 in cellular protrusions led us to investigate FA dynamics in these cancer cell lines. Cells were transiently transfected with FAK-GFP, and FRAP of FAK-GFP in FAs was determined as previously described (23, 30, 31). Integrin-mediated adhesion sites show molecular and functional heterogeneity so that only peripheral FAs were photobleached, and the rate of fluorescence recovery in the bleached region followed over time (Fig. 4B and C). Increased FAK exchange between FA and cytosolic pools, reflected in the higher mobile fraction, was consistently detected in pY14Cav1-negative MDA-435, DU145, and HT29 cells compared with pY14Cav1-positive MDA-231, PC3, and HCT116 tumor cell lines (Fig. 4D; Supplementary Table S1). Treatment of cells with the ROCK inhibitor Y27632 and the Src inhibitor PP2, under conditions that reduce Cav1 tyrosine phosphorylation, selectively affect FAK exchange in pY14Cav1 expressing MDA-231, PC3, and HCT116 tumor cell lines. Y27632 and PP2 did not affect FAK exchange in MDA-435, DU145, and HT29 tumor cells that do not express elevated levels of pY14Cav1 (Fig. 4D; Supplementary Table S1).

Transfection of MDA-435 cells with Cav1-mRFP, but not Cav1Y14F-mRFP, results in FAK stabilization in FAs (23). Similarly, transfection of either MDA-435 or DU145 cells with Cav1-mRFP or the phosphomimetic Cav1Y14D-mRFP, but not Cav1Y14F-mRFP, showed increased FAK stabilization in FAs (Fig. 5A). Importantly, whereas Cav1 expression rendered FAK exchange sensitive to PP2 and Y27632, in cells expressing the phosphomimetic Cav1Y14D, FAK-GFP stabilization in FAs was no longer sensitive to Src or ROCK inhibition (Fig. 5A; Supplementary Table S1). This provides direct evidence of a role for tyrosine phosphorylation of Cav1 in Src-dependent and ROCK-dependent regulation of FAK stabilization in FAs.

A stable lentiviral-infected Cav1 shRNA MDA-231 cell line expresses reduced Cav1 levels (24) and shows an increased FAK-GFP mobile fraction in FAs that is no longer sensitive to PP2 and Y27632 treatment (Fig. 5B; Supplementary Table S1). Furthermore, both ROCK and Src siRNA increased the rate of FAK exchange in MDA-231 cells, but not FAK exchange in Cav1 shRNA MDA-231 cells (Fig. 5C; Supplementary Table S1). pY14Cav1 expression is therefore associated with ROCK-sensitive and Src-sensitive FA dynamics and tumor cell migration.

Expression of Cav1 is a critical determinant of Src-sensitive and ROCK-sensitive tumor cell migration and invasion. We subsequently generated stable MDA-435 cell lines expressing Cav1-mRFP, Cav1Y14F-mRFP, or, as a control, dsRed. These cells expressed Cav1-mRFP at similar levels to endogenous Cav1 and overexpression of WT Cav1, but not Cav1Y14F or dsRed, results in pY14Cav1 expression (Fig. 6A). FAK stabilization in FAs was significantly increased in stable Cav1-mRFP expressing MDA-435 cells relative to both dsRed and Cav1Y14F-mRFP expressing cells; FAK stabilization in FAs of the stable Cav1Y14F-mRFP cells was significantly reduced relative to dsRed-transfected MDA-435 cells, and these cells presented a more spread morphology relative to the other two cell lines (Supplementary Fig. S2A and B; Supplementary Table S1). Expression of WT or mutant versions of Cav1 did not significantly influence cell proliferation or colony formation (Supplementary Fig. S2C and D). Interestingly, RhoA-GTP was down-regulated in MDA-231 cells upon Cav1 shRNA silencing and up-regulated in MDA-435 cells transfected with Cav1-mRFP, but not with either Cav1Y14F-mRFP or dsRed (Fig. 6A). pY14Cav1 expression therefore regulates RhoA activation state in these cells.

Furthermore, migration of MDA-435 expressing Cav1-mRFP, but not Cav1Y14F-mRFP, was elevated relative to dsRed transfection controls. Of particular importance, upon expression of WT-Cav1-mRFP in MDA-435 cells, migration of the cells became ROCK-dependent and Src-dependent (Fig. 6B,, left). MDA-231 cells stably expressing Cav1 shRNA showed reduced migration relative to parental MDA-231 cells, and migration of shRNA Cav1-expressing MDA-231 cells was no longer sensitive to ROCK or Src inhibition (Fig. 6B , right).

Using a Matrigel invasion assay, the elevated invasion of MDA-231 cells was reduced by Cav1 silencing by shRNA whereas overexpression of Cav1-mRFP enhanced the invasive capability of MDA-435 cells. dsRed and pY14F-Cav1 expression did not affect invasion of MDA-435 cells (Fig. 6C). Invasion of only MDA-231 and Cav1-mRFP–expressing MDA-435 cells was sensitive to inhibition of ROCK with Y27632 (Fig. 6D). Expression of Cav1, and more specifically its Y14 phosphorylated form, is therefore associated with ROCK and Src stimulation of tumor cell migration and invasion by stabilizing FAK in FAs and thereby increasing FA dynamics. The parallel role for Cav1 in FA dynamics and migration of cell lines of various tissue origin suggests that this may be a general mechanism by which Cav1 can influence tumor cell invasion and metastasis.

The CAV1 gene maps to a tumor suppressor locus (D7S522; 7q31.1) frequently deleted in human carcinomas, including breast cancer (14). Up to 16% of human breast cancers express a CAV1 P132L mutation that correlates with breast tumor progression and acts as a DN for scaffold domain-dependent growth suppression (32, 33). The negative regulatory function of Cav1 can also be overridden by competitive recruitment of epidermal growth factor receptor to the positive signaling environment of the galectin lattice (34). Mechanisms to override negative regulation of growth signaling by Cav1 will allow tumor cell expression of Cav1 but do not explain the close association between Cav1 and poor prognosis identified in this study and others (15, 16, 35–37).

In prostate cancer, Cav1 has long been associated with poor prognosis and poor patient survival (15). In breast cancer, Cav1 is overexpressed in invasive compared with benign and in situ breast tumors and associated with the basal-like phenotype in sporadic and hereditary breast cancer and poor patient outcome (16, 38, 39). Our finding that Cav1 overexpression is an independent predictor of poor disease prognosis for sporadic invasive breast cancer suggests that Cav1 expression is generally associated with aggressive clinical behavior and decreased survival for breast cancer patients. For colon cancer, Cav1 has been reported as overexpressed in tumor compared with normal mucosa and premalignant lesions, but no consistent relationship between Cav1 and patient prognosis or outcome has been reported (40, 41). The association we observed between Cav1 overexpression and the presence of distant metastases is based on overexpression of Cav1 by 3 of 111 (2.7%) M0 and 2 of 9 (22.2%) M1 patient cancers (P = 0.045) and suggests a role for Cav1 in advanced stages of colon cancer. For rectal cancer, Cav1 was a significant predictor of reduced survival and its expression was associated with recurrence. These observations suggest that Cav1 expression plays distinct roles in the tumor progression of these and other cancer types that may affect the relationship of Cav1 expression to patient outcome. Cav1 is a multifunctional protein, and our observations, as well as those in the current literature, highlight the varied contribution of Cav1 to tumor biology, but nevertheless suggest that Cav1 expression does play an important role in cancer progression and metastasis.

The demonstration here that tyrosine phosphorylated Cav1 is an effector of Rho/ROCK signaling in FA dynamics and tumor cell migration and invasion identify a novel function for Cav1 in tumor progression and metastasis. pY14Cav1 has been localized to the major sites of tyrosine-kinase signaling, FAs, where it generates a docking site for SH2-domain containing proteins, such as Grb7 (42), as well as for the COOH terminal Src kinase Csk that is able to down-regulate Src activity via phosphorylation (43). Cav1 also functions as a membrane adaptor that, when phosphorylated upon integrin ligation in FAs, promotes cell signaling and actin reorganization (20, 21, 44–47). The ability of Cav1 to regulate domain formation in FAs (22) is consistent with the ability of pY14Cav1 to stabilize FAK within FAs, limiting exchange of FAK between FAs and cytosol (23). Reducing FAK exchange enables recruitment of effectors that lead to FA disassembly and turnover, thereby promoting cell migration (30, 31). However, the clear association of Cav1 with integrin and its ability to regulate functional domain organization within FAs define an important role for pY14Cav1 in FA dynamics.

Rho/ROCK signaling, therefore, functionally interacts with pY14Cav1-mediated FAK exchange defining a novel interaction between these metastasis-related processes. Importantly, RhoC expression correlates with Cav1 and Cav2 expression in inflammatory breast cancer (39). Enrichment of pY14Cav1 in pseudopodial domains of tumor cells and the Y14 dependence of Cav1 distribution to tumor cell protrusions localize this interaction to protrusive domains of tumor cells. Activated Rho has been localized to tumor cell protrusions (5–7), and Cav1 polarization to the front of endothelial cells migrating through filter pores is dependent on an intact Y14 residue (48). Cav1 has been reported to interact directly with both RhoA and ROCK and to promote Rho signaling through inhibition of Src-p190RhoGAP signaling to Rac (19, 49, 50). The ability of Rho/ROCK signaling to promote Cav1 tyrosine phosphorylation in the tumor cell lines studied here is indicative of a feedback loop that maintains local Rho activation and pY14Cav1 expression in tumor cell protrusions.

These studies suggest that Cav1 is a critical effector of Rho/ROCK-dependent and Src-dependent tumor cell migration and invasion through regulation of FA dynamics. MDA-231 cells, shown here to present elevated pY14Cav1, Rho activation, and Rho/ROCK-dependent migration, exhibit a proteolytic, mesenchymal mode of migration (28). Regulation of tumor cell invasion by Rho/ROCK signaling may therefore involve Cav1-dependent FA turnover in protrusive tumor cell domains of cells invading via a mesenchymal migratory mode in addition to matrix deformation and proteolysis-independent cell invasion (10, 11). In addition, Rho/ROCK signaling regulates mRNA translocation to tumor cell protrusions, the invasive response to hypoxic conditions and invadopodia formation (5, 8, 29), suggesting that it may play multiple important roles in tumor cell migration and invasion. The critical role described here for pY14Cav1, as a regulator of Rho activation and of Src-dependent and ROCK-dependent tumor cell migration and invasion, provides a mechanistic explanation for the close association between Cav1 expression and poor survival and distant metastasis in various human tumors.

No potential conflicts of interest were disclosed.

Grant support: Cancer Research Society Strategic Program in Genomics and Proteomics of Metastatic Cancer and Canadian Institutes for Health Research grant MOP-64333.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.