ß1 integrin binding phosphorylates ezrin at T567 to activate a lipid raft signalsome driving invadopodia activity and invasion

Abstract

Extracellular matrix (ECM) degradation is a critical process in tumor cell invasion and requires matrix degrading protrusions called invadopodia. The Na(+)/H(+) exchanger (NHE1) has recently been shown to be fundamental in the regulation of invadopodia actin cytoskeleton dynamics and activity. However, the structural link between the invadopodia cytoskeleton and NHE1 is still unknown. A candidate could be ezrin, a linker between the NHE1 and the actin cytoskeleton known to play a pivotal role in invasion and metastasis. However, the mechanistic basis for its role remains unknown. Here, we demonstrate that ezrin phosphorylated at T567 is highly overexpressed in the membrane of human breast tumors and positively associated with invasive growth and HER2 overexpression. Further, in the metastatic cell line, MDA-MB-231, p-ezrin was almost exclusively expressed in invadopodia lipid rafts where it co-localized in a functional complex with NHE1, EGFR, ß1-integrin and phosphorylated-NHERF1. Manipulation by mutation of ezrins T567 phosphorylation state and/or PIP2 binding capacity or of NHE1s binding to ezrin or PIP2 demonstrated that p-ezrin expression and binding to PIP2 are required for invadopodia-mediated ECM degradation and invasion and identified NHE1 as the membrane protein that p-ezrin regulates to induce invadopodia formation and activity.

Figure 1. Analysis of ezrin and p-(T567) ezrin expression and localization in human normal and breast cancer.

(A) representative immunolocalization (IHC) of total ezrin and p-ezrin in a normal breast and in breast tumor sections of increasing Grade from left to right panels, bar equals 10µm. The increased ezrin and p-ezrin expression in the large disorganized, breast tumor lobules (t) displays diffuse distribution in the cells of the lobule while normal, organized breast lobules (n) have a strictly apical distribution. (B) Histogram showing that total ezrin in tumor samples increases its relative membrane localization as Grade increases. (C) Histogram showing that relative p-ezrin expression in tumor samples increases as Grade increases. (D) Immunofluorescence was directly performed for HER2 (green) and p-ezrin (red) in a breast tumor section. The merge and the magnification of the area in the white square shows that there was a high amount of co-localization of the two proteins in the membrane. bar = 10µm.

Figure 2. p-(T567) ezrin is preferentially expressed in invadopodia.

Invadopodia protein expression on 2% cross-linked gelatin layers. (A) Confocal immunofluorescence micrograph of MDA-MB-231 cells cultured on cross-linked rhodamine (TRITC)-conjugated gelatin. Matrix degradation is visualized as loss of red fluorescence background and cells were stained with anti-cortactin antibody (blue) and anti-p-ezrin antibody (green). There was widespread, focalized zones of degradation and cortactin and p-ezrin were both localized and often co-localized in digested areas as indicated by white arrow in both representative axial (XY) and sagittal (XZ) sections. Representitive IF of 3 independent experiments. Scale bars = 10 µm. (B & C) To better visualize invadopodia entrapped in the ECM, the cell bodies of cells incubated for 6 hr on a layer of 2% cross-linked gelatin were removed and separated into cytosol and plasma membrane while entrapped invadopodia were extracted from the gelatin as described in Methods. Proteins of the three fractions were assayed in Western Blot analysis. The upper panel of (B) shows the expression of p-ezrin with a specific antibody and with an antibody for all the forms of phosphorylated ERM proteins and an antibody for total, non-phosphorylated ezrin, while the lower panel displays a histogram of the relative expression of p-ezrin or p-ERM to total ezrin in the three fractions, n= 12. p<0.001 with respect to expression in the invadopodia. (C) expression of a series of proteins to determine their relative expression in the fractions. (D) Using the above scraping protocol on coverslips, immunofluorescence was directly performed for p-ezrin (green) and actin (red) in the invadopodia that remained in the cross-linked gelatin. Representitive IF of 4 independent experiments. Axial (XY) and sagittal (XZ) sections show a high co-localization of p-ezrin and actin in invadopodia. Scale bars = 2 µm.

Figure 3. p-ezrin is expressed and co-localized with NHE1 in invadopodia of cells on Matrigel.

To better visualize invadopodial focal digestion and protein localization in Matrigel, we developed an in situ zymography technique using the quenched fluorescent substrate, DQ-Green-BSA, such that a quantifiable fluorescence is released only upon digestion of the matrix. Cells seeded on Matrigel were allowed to digest the fluorigenic substrate and after fixation cells were assayed in immunofluorescence. The images show (A) cortactin and p-ezrin and (B) NHE1 and p-ezrin immunolocalization at BSA proteolytic spots (green) in axial planes taken at the bottom of the cells (XY). For each of the two fields, XZ zoomed sections of the above representative regions of interest (red box) are shown at the side. As can be seen in green, low levels of basal, diffuse digestion were observed under and around the cells together with more restricted areas of high levels of focal digestion. Representative IFs of 5 independent experiments. Scale bar = 10 µm. Right panels: XZ plane and relative magnification of indicated area show the high co-localization of cortactin and NHE1 with p-ezrin and proteolysis in RGB analysis. Lower panels: ICA analysis plots using the JACoP image analysis plugin in ImageJ of cortactin/p-ezrin and NHE1/p-ezrin shows their highly significant co-localization. Average ICQ values for all the experiments are presented in the text.

Figure 4. p-ezrin binds NHE1, p-NHERF1, PKA and ß1 integrin/EGFR in a ternary complex in tumor cells.

(A) MDA-MB-231 cells were seeded for 24 hr on 2% gelatin and cytosol, cell body membrane and invadopodia fractions were separated as above. Each cell membrane or invadopodia fraction was then immunoprecipitated with anti-NHE1, anti-p-ezrin or anti-ezrin and the precipitated immunocomplex was probed for the expression of NHE1, p-ezrin, NHERF1 and RIIß subunit of PKA by Western Blot. (B) the fractions were immunoprecipitated with anti-p-ezrin and the precipitated immunocomplex was probed for the expression of ß1 integrin and EGF receptors by Western Blot. Protein input for each protein was measured by Western Blot of the total input with the same antibodies. Representitive blots of 4 independent experiments.

Figure 5. NHE1 and p-ezrin are localized in the same lipid raft fraction in the invadopodia but not in the membrane compartments and lipid raft cholesterol is necessary for invadopodia function.

(A) Raft (pellet) and nonraft (supernatant, Supernat) fractions were separated from the above membrane and invadopodia fractions by incubating the fractions (100 µg protein) at 4°C with lysis buffer containing 1% of Lubrol. Insoluble material was collected into pellets (centrifugation at 100,000 X g for 1 hour), and equal amounts of the resuspended pellet (P) and the supernatant (SN) were analyzed by Western Blotting for NHE1, p-ezrin, total ezrin, EGFR and ß1-integrin. Quality of separation was determined by immunoblotting for flotillin-1, a marker for lipid rafts. Representative blots are shown (n = 3). To determine the effect of cholesterol depletion on invadopodia function, MDA-MB-231 cells were incubated for 30 minutes at 37°C in the presence (MβCD-treated) or absence (untreated) of 0.5% MβCD as described in Methods and invadopodia-dependent ECM digestion was analyzed in confocal microscopy for a series of individual cells as described in Methods; (B) typical images of digestion for 6 hrs and overnight in cells treated or not. (C) histogram of a series of experiments performed as in (B). Mean ± S.E.M., n=3, ***p<0.001 for focal proteolysis compared to the control cells. (D) NHE1 immunofluorescence of vehicle or MβCD treated cells demonstrating a change in transporter distribution but not total expression. bar = 10 µm.

Figure 6. Ezrin phosphorylation and binding to PIP2 and NHE1 binding to both ezrin and PIP2 are necessary for invadopodia proteolytic activity.

(A) To examine the role of ezrin T567 phosphorylation and/or binding to PIP2 in invadopodial-dependent focal digestion of the ECM, MDA-MB-231 cells were (A) transfected with empty cDNA (Control) or plasmids contain ezrin cDNA mutated in the T567 site to be either phosphodead (pd) or phosphomimic (pmim) or mutated in the ezrin PIP2 binding site such that it no longer can bind PIP2 (PIP2-) or treated with 1µM of the inhibitor of ezrin T567 phosphorylation, NSC668394 (NSC). (B) To examine the role of NHE1 binding to either ezrin or PIP2 in invadopodia proteolysis, MDA-MB-231 cells were transfected with an NHE1 mutant lacking the ability to bind to ezrin (KR/A 556-564-NHE1-HA) or to bind to PIP2 (KR/A 513-520-NHE1-HA). Two days after transfection, cells were plated on Matrigel with DQ-Green BSA and, 24 hr later, ECM digestion was analyzed in confocal microscopy for a series of individual cells as described in Methods. Mean ± S.E.M., n=4, ***p<0.001 for focal proteolysis compared to the control cells.

Figure 7. Ezrin phosphorylation and binding to PIP2 are necessary for invasion.

To examine the role of ezrin T567 phosphorylation and/or binding to PIP2 in invasive capacity, MDA-MB-231 cells were transfected with empty cDNA (Control) or plasmids containing ezrin cDNA mutated in the T567 site to be either phosphodead (pd) or phosphomimic (pmim) or mutated in the ezrin PIP2 binding site such that it no longer can bind PIP2 (PIP2-). Two days after transfection, a quantitative measure of the degree of in vitro invasion of MDA-MB-231 cells was measured as the ability to traverse a 8 µm polycarbonate membrane coated with 5 µg matrigel (Chemicon Int., Livermore, CA) as previously described (Cardone et al., 2005). The fluorescent samples were read in a fluorescence plate reader at 480/520 nm (Cary Eclipse Fluorescence Spectrophotometer, Varian). Mean ± S.E., n=5, ***p<0.001 and n.s. is not significant.