Invadopodia biogenesis is regulated by caveolin‐mediated modulation of membrane cholesterol levels

Abstract

Invadopodia are proteolytically active protrusions formed by invasive tumoural cells when grown on an extracellular matrix (ECM) substratum. Clearly, invadopodia are specialized membrane domains acting as sites of signal transduction and polarized delivery of components required for focalized ECM degradation. For these reasons, invadopodia are a model to study focal ECM degradation by tumour cells. We investigated the features of invadopodia membrane domains and how altering their composition would affect invadopodia biogenesis and function. This was achieved through multiple approaches including manipulation of the levels of cholesterol and other lipids at the plasma membrane, alteration of cholesterol trafficking by acting on caveolin 1 expression and phosphorylation. We show that cholesterol depletion impairs invadopodia formation and persistence, and that invadopodia themselves are cholesterol‐rich membranes. Furthermore, the inhibition of invadopodia formation and ECM degradation after caveolin 1 knock‐down was efficiently reverted by simple provision of cholesterol. In addition, the inhibitory effect of caveolin 3^DGV expression, a mutant known to block cholesterol transport to the plasma membrane, was similarly reverted by provision of cholesterol. We suggest that invadopodia biogenesis, function and structural integrity rely on appropriate levels of plasma membrane cholesterol, and that invadopodia display the properties of cholesterol‐rich membranes. Also, caveolin 1 exerts its function in invadopodia formation by regulating cholesterol balance at the plasma membrane. These findings support the connection between cholesterol, cancer and caveolin 1, provide further understanding of the role of cholesterol in cancer progression and suggest a mechanistic framework for the proposed anti‐cancer activity of statins, tightly related to their blood cholesterol‐lowering properties.

Introduction

Invasive tumoural or transformed cells grown on an extracellular matrix (ECM) substratum extend proteolytically active protrusions into the matrix from their ventral surfaces; these protrusions have been termed invadopodia [1]. Invadopodial protrusions are actin‐based structures enriched in integrins, soluble and membrane proteases, including matrix metalloproteases (MMPs), and actin‐associated proteins [reviewed in 2, 3, 4]. Invadopodia are also hot spots of tyrosine‐phosphorylation [5] suggesting a very intense local signalling activity. These features define them as powerhouses for the focal degradation of the ECM, made possible by a tight integration between the signalling, membrane trafficking and cytoskeletal machineries. ECM degradation at invadopodia is organized in a highly focalized fashion. Indeed, the concentration of certain proteases (e.g. the membrane type‐1 MMP, MT1‐MMP) and integrins at invadopodia and the fact that the Golgi complex is often found to be close to invadopodia [6], very much akin to the Golgi complex reorientation occurring in motile cells [7], suggest that specific sorting processes might be taking place. Although their molecular definition is still incomplete, these observations characterize invadopodia as specialized membrane domains for focalized ECM degradation.

Signal transduction, local cytoskeletal remodelling events and polarized trafficking of proteins have been associated with specific membrane microdomains, typically enriched in cholesterol and sphingolipids, generally referred to as lipid rafts [8], cholesterol‐enriched membrane microdomains (CEMM) or detergent‐resistant membranes, due to their reduced solubility in mild non‐ionic detergents. These membrane subdomains represent the ‘liquid ordered’ phase in the bulk plasma membrane. CEMM have proved difficult to visualize, possibly due to their dynamic features and small size, so that their very existence has been questioned [9, 10]. Functional data, however, support their existence as domains where proteins with affinity for specific lipid combinations (including cholesterol and sphingolipids) cluster together dynamically to promote or inhibit the interaction between different classes of proteins, thus spatially and temporally compartmentalizing many cell functions.

The integrity of CEMM has been shown to have a role in tumour‐promoting pathways, leading to enhanced cell proliferation and cell survival [11]. Furthermore, the incidence of certain solid tumours has been related to diets rich in saturated fatty acids and cholesterol [12]. Also, cholesterol accumulates in solid tumours [reviewed in 13], and fundamental cholesterol synthesis feedback regulation mechanisms are lost in prostate and other cancer types, leading to enhanced cholesterol synthesis, low‐density lipoprotein (LDL) receptor up‐regulation [14, 15] and CEMM enrichment [16].

Cellular cholesterol homeostasis and availability at the plasma membrane are tightly controlled by proteins of the caveolin family [17]. They were first discovered as the major component of caveolae, a flask‐shaped subset of CEMM domains on the plasma membrane. The three known caveolins are peculiar hairpin‐structured integral membrane proteins with cytosolic N‐ and C‐termini and lacking an extracellular domain. Caveolins 1 and 3, expressed in most non‐muscle tissues and striated muscle cells respectively, are required for the formation of caveolae. They are closely related to each other, so that caveolae induced by heterologous expression of the two isoforms are morphologically indistinguishable [18]. In contrast, caveolin 2 displays considerably lower sequence homology, is tightly co‐expressed with caveolin 1 and appears to play a role in caveolae formation only in particular cell types [19]. The caveolins are considered to be multi‐functional players involved in vesicle trafficking [20, 21], signal transduction [22], and more broadly, in cancer [23]. Caveolin is one of the few proteins that binds cholesterol tightly and specifically [24] and is a major mediator of cholesterol transport to the plasma membrane [25].

Hence, considering the possible link between cholesterol levels, caveolin expression and cancer, and the role of cholesterol in polarized trafficking and signalling, we investigated the features of invadopodia membrane domains and how altering their composition would affect invadopodia biogenesis and function. This was achieved through a number of approaches including direct manipulation of the levels of cholesterol and other lipids at the plasma membrane, alteration of cholesterol trafficking by acting on caveolin 1 expression and phosphorylation.

The evidence presented herein represents novel insight toward the elucidation of invadopodia structure and functionality, indicates caveolin 1 as a master regulator of invadopodia biogenesis through the modulation of lipid domains, suggests a new scenario to explain the potential anti‐cancer properties of statins, linked to their cholesterol‐lowering effects and also highlights cholesterol‐rich domains as a pharmacological target to inhibit invasion and metastasis.

Materials and methods

Cell culture

Human melanoma A375MM cells were cultured in DMEM/F‐12 (1:1) (Invitrogen, Carlsbad, CA, USA) containing 10% foetal calf serum (FCS). Cells were grown at 37°C in a humidified atmosphere containing 5% CO₂ as previously described [6].

Antibodies, reagents and constructs

Alexa Fluor 633‐, 546‐ and 488‐conjugated secondary antibodies and phalloidin, as well as Alexa Fluor 555‐conjugated cholera toxin subunit B, were from Molecular Probes (Leiden, The Netherlands). Peroxidase‐ conjugated secondary antibodies were from Calbiochem (San Diego, CA, USA). Anti‐cortactin monoclonal mouse antibody was from Upstate Biotechnology (Lake Placid, NY, USA). Anti‐dynamin 2 monoclonal antibody was from BD Transduction Laboratories (Lexington, KY, USA). Polyclonal antibodies against glyceraldehyde 3‐phosphate dehydrogenase and glucosylceramide synthase were purchased from Biogenesis (Kingston, NH, USA) and Exalpha Biologicals, Inc. (Maynard, MA, USA), respectively. Phosphorylated Src was detected with a rabbit polyclonal antibody (Y418; Biosource International, Belgium). Mouse monoclonal anti‐HA was from Covance (Denver, PA, USA). HA‐tagged caveolin 3^WT, caveolin 3^DGV and caveolin 3^KSY were a kind gift of Robert G. Parton (University of Queensland, Australia). GFP‐tagged rat caveolin 1^WT, caveolin 1^Y14D and caveolin 1^Y14F were a kind gift of Mark McNiven (Mayo Clinic College of Medicine, Rochester, MN, USA). Lovastatin, mevalonate, methyl‐β‐cyclodextrin, cholesterol, filipin, fumonisin B1 as well as Optiprep Density Gradient Medium were purchased from Sigma (St Louis, MO, USA). BB94 (British Biotech, UK) was solubilized in 70% ethanol to 1 mM final concentration and kept in aliquots at –20°C until use.

Transfection

Cells were plated at 50% confluence in 6‐well plates. The next day, they were washed twice with DMEM/F‐12 without FCS and incubated in 0.5 ml DMEM/F‐12 without FCS containing 2–4 μg DNA and 8 μl TransFast (Promega, Madison, WI, USA) at 37°C. After 1 hr, 1.5 ml of complete medium were added. Experiments involving transient overexpression of exogenous proteins were usually performed 24 hrs after transfection.

Invadopodia labelling and immunofluorescence

Invadopodia can be identified by immunofluorescence microscopy as dots enriched in actin, cortactin, tyrosine‐phosphorylated proteins and dynamin 2, colocalizing with the underlying degradation patches, represented by dark holes in the fluorescent gelatin. To this end, cells were fixed in 4% paraformaldehyde for 15 min., permeabilized in PBS containing 0.02% saponin, 0.2% bovine serum albumin (BSA) and 50 mM NH₄Cl, incubated with phalloidin or the primary antibodies of interest for 1 hr. In the latter case, samples were further incubated with fluorophore‐conjugated secondary antibodies for 45 min. Finally, coverslips were mounted in Mowiol (Calbiochem). Experiments were observed using either an LSM 510 laser scanning confocal microscope (Carl Zeiss, Germany) or a T.I.L.L. Photonics video microscope (T.I.L.L. Photonics GmbH, Germany).

ECM degradation assay

Fluorescent matrix‐coated coverslips were prepared, and the assay carried out, as described previously [6, 26]. Briefly, thin layers of fluorescein‐ or rhodamine B (Sigma‐Aldrich)‐conjugated porcine gelatin (Sigma‐Aldrich) were placed on coverslips, cross‐linked with 0.5% glutaraldehyde for 15 min. on ice, washed three times with PBS and incubated for 3 min. at room temperature with 5 mg/ml NaBH₄. Finally, after three washes with PBS and 5‐min. incubation in 70% ethanol, coverslips were maintained with complete medium for 1 hr at 37°C before cell plating. From here on, cross‐linked gelatin, prepared as described above, will be referred to simply as ‘gelatin’. Cells (1.5 × 10⁵) were plated on gelatin according to the previously published invadopodia synchronization protocol with some modifications [27]. Briefly, A375MM cells were plated in the presence of 5 μM BB94 to block invadopodia formation. After 16 hrs, BB94 was washed out to allow synchronous invadopodia formation. Samples were fixed after 3 hrs and stained as described in the previous paragraph. Areas of degradation were quantified considering 30 random fields per each condition using the public domain software ImageJ v.1.34 from the NIH. The total degradation area was then normalized for the number of cells and the effect of the treatment was reported as a percentage of the control.

RNA interference and rescue experiments

A375MM cells were transfected with the siGenome smart pool reagents specific for human caveolin 1 and a set of four duplexes for glucosylceramide synthase (Dharmacon, Lafayette, CA, USA) using Oligofectamine (Invitrogen) according to the manufacturer’s instructions. Silencing efficiency was evaluated by Western blotting 72 hrs after transfection. Both treated and control cells were plated on fluorescent matrix‐coated coverslips 53 hrs after siRNA treatment and the degradation evaluated as indicated above. Rescue experiments were performed by transfecting cells 48 hrs after siRNA‐treatment targeting human caveolin 1, with GFP‐tagged rat caveolin 1^WT, caveolin 1^Y14D and caveolin 1^Y14F. Cells were then plated on gelatin and processed as indicated above. To exclude off‐target effects we also tested a pool of four non‐targeting siRNA duplexes. Quantification of phosphorylated Src levels was performed by estimating fluorescence with the LSM510–3.2 software (Zeiss), using untransfected cells in the same field as a control. At least 20 random fields were quantified per each sample as a ratio between fluorescence and total cell area. The results are expressed as a percentage of pSrc intensity in untreated cells.

Cholesterol depletion, re‐addition and quantification

Cholesterol depletion was performed as previously published, with some modifications [32]. Briefly, cells pre‐incubated for 16 hrs with 4 μM lovastatin and 0.25 mM mevalonate were treated with 10 mM methyl‐β‐cyclodextrin for 40 min. in serum‐free medium, in combination with the above‐indicated concentration of lovastatin and mevalonate. After methyl‐β‐cyclodextrin washout, complete medium was re‐added, and cells were allowed to degrade for 3 hrs. Control cells were left untreated and processed in parallel in the absence of methyl‐β‐cyclodextrin. Samples were then processed and the effect of the treatment quantified as a ratio between cells bearing invadopodia and the total number of cells considering at least 30 random fields per each condition. For experiments aimed at evaluating the effect of methyl‐β‐cyclodextrin on invadopodia biogenesis, degradation was quantified as described above.

Cholesterol re‐addition was performed as previously described [28] with some modifications. Briefly, a cholesterol methyl‐β‐cyclodextrin mix at the final concentration 0.04 mg/ml and 1%, respectively, was prepared by adding 20 μl of cholesterol solution (20 mg/ml in ethanol) to 10 ml of DMEM/F‐12 containing methyl‐β‐cyclodextrin 1%, while vortexing at 40°C. To replenish cholesterol, cells were treated with the cholesterol/ methyl‐β‐cyclodextrin mix for 1 hr at 37°C in the presence of BB94. After BB94 washout, complete medium was re‐added for 3 hrs. Samples were then processed and the degradation quantified as described above.

Cholesterol was quantified with the Amplex Red Cholesterol assay kit (Invitrogen) according to the manufacturer’s instructions, by using the same amount of cell lysate in RIPA buffer (50 mM Tris–HCl, 150 mM NaCl, 1% NP‐40, 0.5% sodium deoxycholate, 2 mM sodium fluoride, 2 mM EDTA, 0.1% SDS) per sample. To stain plasma membrane cholesterol, cells were fixed for 1 hr with 3% paraformaldehyde in PBS, incubated for 10 min. with 1.5 mg/ml glycine in PBS. Samples were then labelled with 0.05 mg/ml filipin in PBS with 10% FCS for 2 hrs.

Optiprep density gradients

Optiprep gradient analysis of Triton X‐100 (TX100)‐insoluble material was performed using previously published protocols with some modifications [29]. Briefly, A375MM cells were grown to 80% confluence in 100 mm dishes pre‐coated with un‐conjugated gelatin in the presence or absence of BB94. BB94 was washed out 16 hrs before, lysis. Cells were washed in PBS and lysed with 1% TX100 in TNE buffer (50 mM Tris‐HCl, 150 mM NaCl, 2 mM EDTA, pH 8.0) for 10 min. on ice. Lysates were scraped from dishes, adjusted with Optiprep to 35%, and placed at the bottom of a centrifuge tube. A discontinuous Optiprep gradient (5–30% in TNE) was layered on top of the lysates and the samples were ultracentrifuged at 120,000 ×g for 16 hrs in a swinging‐bucket rotor. One ml fractions were harvested from the top. Proteins were precipitated from each fraction with tricloroacetic acid, washed twice with acetone, dissolved in sample buffer and separated by SDS‐PAGE (10%). Proteins were then electro‐blotted and probed with anti‐cortactin, anti‐dynamin 2 and anti‐caveolin antibodies. Optiprep gradient analysis was also performed on invadopodia‐enriched fractions. To prepare invadopodia‐enriched fractions, a published protocol was followed with some modifications [30]. In detail, 16 hrs after plating, A375MM cells were washed three times with PBS plus 0.5 mM MgCl₂ and 1 mM CaCl₂ (aPBS), and two with a hypo‐osmotic buffer (5‐fold diluted PBS plus 0.5 mM MgCl₂ and 1 mM CaCl₂) (bPBS). Cells were then incubated 15 min. in bPBS containing a protease inhibitor mix. Cell bodies were sheared away from adherent cells by using an L‐shaped glass Pasteur pipette. The remnants embedded in the gelatin‐coated dish were washed twice with aPBS and scraped away with TNE containing 1% TX100. The so obtained invadopodia‐enriched fraction was clarified by centrifugation at 18,000 ×g for 15 min. at 4°C and subjected to Optiprep gradient analysis as described above.

In situ isolation of detergent‐resistant membranes

A375MM cells plated on gelatin for 16 hrs. Cells were then washed twice with PBS containing 1 mM CaCl₂ and 1 mM MgCl₂ (cPBS) and treated with 1% TX100 containing buffer (PIPES 10 mM pH 6.8, 100 mM KCl, 2.5 mM MgCl₂, 1 mM CaCl₂, 300 mM sucrose, protease inhibitors) for up to 5 min. After washing with cPBS, the so obtained TX100‐resistant fraction was stained with anti‐cortactin antibodies as described above. To label GM1 ganglioside, fixed cells were incubated with Alexa Fluor 555‐conjugated cholera toxin subunit B (10 μg/ml in PBS with 0.1% BSA) for 15 min.

Correlative light‐electron microscopy (CLEM)

CLEM was performed as previously described [31], with some modifications. Briefly, A375MM cells were grown on gelatin‐coated MatTek (Ashland, MA, USA) glass bottom Petri dishes. After 16 hrs, samples were fixed with 0.05% glutaraldehyde plus 4% paraformaldehyde in 0.2M HEPES (pH 7.4) for 5 min. and then with 4% paraformaldehyde in the same buffer for 30 min. Next, cells were incubated in 0.02% saponin‐containing blocking solution (PBS with 0.2% BSA and 50 mM NH₄Cl); samples were then incubated with phalloidin for 1 hr and analysed using a laser scanning confocal microscope to identify the cell bearing the structure of interest within the grid coordinates.

After washing, samples were incubated with anti‐caveolin 1 antibody and revealed with Gold Enhancement‐EM (Nanoprobes, Stony Brook, NY, USA) according to the manufacturer’s instructions. Samples were processed for conventional EM with 1% OsO4 plus 1.5% potassium ferrocyanide in 0.1 M cacodylate buffer (pH 7.3) for 3 hrs at room temperature in the dark, followed by incubation with 0.5% thiocarbohydrazide in 0.1 M cacodylate buffer (pH 6.8) 5 min. They were placed again in 1% OsO4 plus 1.5% potassium ferrocyanide in 0.1 M cacodylate buffer (pH 7.3) for 20 min. followed by incubation with 1% tannic acid for 1 hr. Then samples were dehydrated, embedded in epoxy resin and polymerized for at least 24 hrs. Coverslips were dissolved with 40% HF and samples extensively washed with buffer. Finally, serial sections of the cell of interest were produced parallel to the substrate. One hundred‐nanometer serial sections were collected on slot grids covered with Formvar‐carbon supporting film and examined with a Tecnai 12 electron microscope at 200 kV (FEI/Philips Electron Optics) equipped with a slow‐scan CCD camera. The images collected by confocal and electron microscopy were aligned with Adobe Photoshop and the structure of interest was identified on the basis of grid coordinates.

Data analysis

Each experiment was repeated at least three times. Where images are shown, they represent typical results. Error bars are standard deviations of the mean.

Results

Cholesterol and sphingolipids as key components of invadopodia

To investigate the role of cholesterol at invadopodia, we employed a procedure based on the use of a lovastatin/mevalonate mix, combined with the short‐term action of methyl‐β‐cyclodextrin (MCD) [modified from 32]. Lovastatin decreases total cellular cholesterol by blocking the mevalonate pathway, whereas MCD extracts cholesterol from the plasma membrane. This procedure has been shown to ensure successful cholesterol depletion without affecting cell viability, depending on the cell type. The addition of small amounts of mevalonate helps to reduce the secondary effects of statins on the synthesis of non‐sterol products [32].

First, we examined the consequence of acute cholesterol extraction on pre‐formed invadopodia. A375MM cells, plated on gelatin and incubated for 16 hrs with or without the lovastatin/mevalonate combination, exhibited active invadopodia visualized as actin dots co‐localizing with the underlying degradation patches (Fig. 1). When lovastatin/mevalonate‐treated cells were subjected to cholesterol extraction with MCD, invadopodia disappeared in 50% of cells as compared to control cells (Fig. 1 and Fig. S1E, F). Since the plasma membrane can be continuously replenished with newly synthesized cholesterol from the endoplasmic reticulum, and because lovastatin/mevalonate alone did not appear to affect invadopodia performance, we chose to always precede cholesterol extraction with lovastatin/mevalonate treatment to reduce available intracellular pools of cholesterol, and to use this as a control in the following experiments.

Figure 1.

Cholesterol depletion disrupts pre‐formed invadopodia. A375MM cells were plated on gelatin for 16 hrs with 4 μM lovastatin/ 0, 25 mM mevalonate. Treated cells were then subjected to acute cholesterol extraction with 10 mM methyl‐beta‐cyclodextrin (MCD) for 40 min., incubated for further 3 hrs and then fixed. Phalloidin staining revealed the presence of intact invadopodia, visualized as actin‐rich dots (arrowheads), co‐localizing with the underlying gelatin degradation patches (arrows) in untreated (A) and lovastatin/mevalonate treated (C) cells. Underlying fluorescent gelatin is shown in B and D, respectively. Following acute cholesterol extraction (E), pre‐formed invadopodia, identifiable by the presence of degradation patches in the substrate (F) formed prior to MCD treatment (arrows), disappeared. Scale bar 10 μm.

Next, we investigated whether plasma membrane cholesterol was also required for invadopodia biogenesis. To this end, cells were plated directly on gelatin in the presence of lovastatin/mevalonate and the broad‐spectrum metalloprotease inhibitor BB94, and incubated for 16 hrs. Invadopodia formation is blocked in the presence of BB94 but upon washout, synchronous invadopodia formation occurs [27]. When cholesterol was extracted with MCD prior to BB94 washout, invadopodia biogenesis was severely impaired, with a consequent 80% reduction of ECM degradation compared to control (Fig. 2A). To verify the specificity of this effect, previously depleted cells were replenished with cholesterol by using a cholesterol/MCD mix prior to BB94 washout. Cholesterol re‐addition partially reconstituted the ability of cells to form functional invadopodia (Fig. 2A). The efficacy of treatments aimed at manipulating cholesterol levels was checked with two different approaches. The first is a colorimetric assay, which allows to assess total cell cholesterol; the second is based on immunofluorescence staining with filipin, which detects plasma membrane cholesterol. Filipin is a polienic cholesterol‐binding antibiotic that mainly labels cell surface cholesterol, when used in non‐permeabilized cells. Not surprisingly, MCD extraction did not dramatically decrease total cell cholesterol levels (Fig. 2B). Conversely, filipin staining revealed a significant reduction of plasma membrane cholesterol following the depletion procedure (Fig. 2C). This suggests that the severe impairment of invadopodia formation consequent to MCD treatment is specifically related to plasma membrane cholesterol depletion, rather to a general cholesterol decrease.

Figure 2.

Cholesterol depletion blocks invadopodia formation. Cells were plated on gelatin with BB94 and left untreated (NT) or treated with the lovastatin/mevalonate mix (L+M). The latter were subjected or not to acute cholesterol extraction with 10 mM MCD (MCD) 40 min. prior to BB94 washout. Samples were fixed after 3 hrs. (A) A degradation assay revealed an 80% decrease in MCD‐treated as compared to lovastatin/mevalonate‐treated cells. Cholesterol re‐addition (Chol) to depleted cells restored the normal phenotype. (B) The total cholesterol content of each sample, relative to untreated cells and measured with the Amplex Red Cholesterol Assay kit, is reported. Error bars are standard deviations of the mean. (C) Filipin staining mainly labelled plasma membrane cholesterol in untreated‐ (NT) and lovastatin/mevalonate‐ (L+M) treated cells. Upon MCD treatment (MCD), plasma membrane staining was essentially abolished and restored following cholesterol re‐addition (Chol). Scale bar 10 μm.

As mentioned above, glycosphingolipids are, together with cholesterol, important lipid constituents of CEMM. Therefore, we investigated whether glycosphingolipid depletion affected invadopodia formation. To this end, we used a two‐pronged approach based on usage of the fungal metabolite fumonisin B1, and siRNA targeting human glucosylceramide synthase. The first, a competitive ceramide synthase inhibitor, impairs synthesis of both glycosphingolipids and sphingomyelin. Glucosylceramide instead, synthase converts ceramide into glucosylceramide, the precursor of glycosphingolipids [33], so that by knocking‐down this enzyme only the level of this subset of sphingolipids is affected. Both treatments impaired invadopodia biogenesis, although with different effectiveness (Fig. 3A–C). This discrepancy could suggest that glycosphingolipids and sphingomyelin are equally required for the degradation of the ECM or alternatively, could be the consequence of inefficient glucosylceramide synthase down‐regulation.

Figure 3.

Inhibition of sphingolipid synthesis affects invadopodia formation. (A) A375MM cells were incubated on gelatin with BB94 in the presence or absence of 25 μg/ml fumonisin B1 for 72 hrs. BB94 was then washed out and cells incubated for further 3 hrs. Treated cells exhibited an 80% reduction of ECM degradation compared to control. (B) A375MM cells were incubated on gelatin and transfected without (mock) or with (GCS‐KD) siRNA targeting glucosylceramide synthase for 72 hrs. Western blot of lysates using antibodies against GCS and GAPDH as a loading control are shown. (C) Quantification of ECM degradation in GCS‐KD cells compared with mock. Error bars are standard deviations of the mean.

Invadopodia are cholesterol‐rich membrane domains

The importance of cholesterol and glycosphingolipids in the establishment of invadopodia, as well as in maintaining the integrity of these structures, would suggest they have the features of CEMM. If this were the case, invadopodia should exhibit a typ ical resistance to solubilization with a mild non‐ionic detergent such as TX100 [8]. To test this, A375MM cells were plated on gelatin and incubated for 16 hrs to allow formation of invadopodia and ECM degradation (Fig. 4A–C). Cells were then incubated with a buffer containing TX100 for up to 5 min., washed and processed as usual for immunofluorescence using cortactin as a bona fide invadopodia marker. In situ cell remnants were readily distinguishable as cortactin‐positive speckles co‐localizing with degradation patches (Fig. 4D–F). The same structures were also efficiently labelled by fluorophore‐conjugated cholera toxin B subunit (Fig. 4G–I), a ligand for GM1 ganglioside, a CEMM component and bona fide marker. The localization of cortactin and GM1 at sites of degradation after detergent extraction also suggests that the cortactin‐positive speckles were genuine membrane remnants and not mere protein aggregates resulting from insolubility in mild detergents.

Figure 4.

Invadopodia are resistant to TX100 extraction. A375MM cells plated on gelatin form invadopodia, visualized as cortactin‐rich dots (A) co‐localizing with underlying matrix degradations, represented by dark holes in a fluorescent background (B). Five minutes following addition of a 1% TX100 containing buffer, cell remnants were visualized as cortactin‐rich (D) and GM1‐positive (G) speckles overlapping with degradation patches (E and H, respectively). Merged images are also shown (C, F, I). Scale bar 10 μm.

As mentioned above, mild non‐ionic detergents hardly solubilize CEMM. Also, when a cell lysate obtained after mild detergent treatment is subjected to a density gradient ultracentrifugation, non‐solubilized membranes (i.e. CEMM) float to the lighter fractions, as opposed to soluble proteins or insoluble protein aggregates (such as cytoskeletal proteins). Hence, to further support our conclusions, we incubated cells on gelatin in the presence or absence of BB94 to obtain monolayers without and with invadopodia, respectively. Cells were then lysed and subjected to Optiprep density gradient centrifugation, the fractions collected and analysed after Western blotting. We observed that, in samples from cells cultured in the continuous presence of BB94 (i.e. without invadopodia), cortactin and dynamin 2 were retrieved mainly in the heavier fractions (Fig. 5A). This reflects the known distribution of these proteins in the cell, with cortactin being mainly actin cytoskeleton‐associated, and dynamin 2 mainly membrane‐associated and cytosolic. When, however, lysates from actively degrading cells (i.e. with invadopodia) were analysed, a small but consistent amount of typical invadopodia components, namely cortactin and dynamin 2, was retrieved in the light density fractions (Fig. 5B). Lipid raft components are usually located within fractions 5–8 in this floatation assay [34]; the use of caveolin 1 as a marker confirmed that these fractions have the properties of bona fide CEMM.

Figure 5.

Cortactin and dynamin 2 partially shift to CEMM when invadopodia are formed. Cells plated on gelatin‐coated Petri dishes without or with BB94 for 16 hrs were lysed in TX100‐containing buffer. Fractions were collected and subjected, together with a sample of the initial lysates (Inp), to Western blotting and probed with anti‐cortactin and anti‐dynamin 2 antibodies. Caveolin 1 was used as a bona fide light‐fraction marker. Two different experiments are shown to present the distribution profile of cortactin and dynamin 2. (A) When invadopodia were not present, cortactin and dynamin 2 were mainly retrieved in the heavier fractions. The corresponding graphs show the percentage of protein per each fraction as a ratio of the total protein. (B) When, instead, cells featured active invadopodia, a small but significant, amount of both cortactin and dynamin 2 floated to the lighter fractions, as highlighted by the relative graphs. (C) Floatation assays were performed also on lysates from invadopodia‐enriched fractions, obtained as described in Materials and Methods. Cortactin was found to be partially but significantly associated with cholesterol‐enriched fractions.

Clearly, when invadopodia are formed, a fraction of dynamin 2 and cortactin becomes CEMM‐associated. It is important to consider that the majority of cortactin and dynamin 2 can be found in the cytosol and other membrane structures and cytoskeletal elements. They are in fact, localized in, but not exclusive to, invadopodia. The possibility remained that the partial shift of cortactin and dynamin 2 to the lighter fractions was not specifically associated to invadopodia. To verify this, we prepared an invadopodia‐enriched cell fraction. This was subjected to the same density gradient centrifugation assay and after Western blotting, analysed for the distribution of cortactin. We found that cortactin was partially retrieved in the cholesterol‐rich CEMM‐containing light fractions (Fig. 5C).

Altogether, this evidence suggests that invadopodia are indeed cholesterol‐rich membranes.

Caveolin 1 is required for cholesterol delivery to the plasma membrane and invadopodia biogenesis

To confirm the importance of cholesterol in regulating invadopodia biogenesis and to gain insight into the molecular mechanism of its regulation, we investigated the role of caveolin 1, a key regulator of cholesterol transport to the plasma membrane [25], in the formation and function of invadopodia. To this end, A375MM cells were treated with siRNA targeting caveolin 1. Both mock and siRNA‐treated cells were plated on gelatin and scored for invadopodia formation and ECM degradation. In cells with reduced caveolin 1 expression levels (up to 85%) (Fig. 6A), there was a 50% reduction in ECM degradation compared to mock (Fig. 6B). The invasive phenotype was efficiently rescued upon re‐expression of rat caveolin 1 in caveolin 1‐depleted cells (Fig. 6B). Filipin staining appeared weaker in caveolin 1‐depleted cells (Fig. 6C) compared to mock, suggesting reduced availability of plasma membrane cholesterol. We thus asked if the role of caveolin 1 in invadopodia biogenesis was connected to its cholesterol trafficking function or to other. To this effect, we knocked down caveolin 1 and incubated cells in the presence of BB94 to allow cell attachment and spreading, without invadopodia formation. Cholesterol was re‐added to caveolin 1‐depleted cells with a cholesterol/MCD mix prior to BB94 washout. Strikingly, cholesterol re‐addition alone sufficed to rescue invadopodia formation and ECM degradation, and was as efficient as rat caveolin 1 re‐transfection (Fig. 6B) (see also Fig. 8A). Successful cholesterol replenishment was confirmed by monitoring plasma membrane cholesterol levels with filipin staining (Fig. 6C).

Figure 6.

Caveolin 1 knock‐down impairs invadopodia formation and the phenotype is efficiently rescued upon cholesterol re‐addition. A375MM cells were treated without (mock) or with (Cav 1‐KD) siRNA targeting caveolin 1 for 72 hrs. (A) RNAi reduced caveolin 1 expression up to 85% as shown in Western blots of lysates using antibodies against Caveolin 1 and glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) as a loading control. (B) Matching quantification of ECM degradation in Cav1‐KD cells indicated a 50% reduction compared to mock. The phenotype was efficiently rescued by overexpression of wild‐type rat caveolin 1 (rCav 1^WT) (see Fig. 6 A) or cholesterol re‐addition (Chol). Error bars represent standard deviation of the mean. (C) Filipin staining revealed a decrease in plasma membrane cholesterol in Cav 1‐KD cells compared to mock, with a recovery to normal levels following cholesterol re‐addition to knocked‐down cells. Scale bar 10 μm.

Figure 8.

Caveolin 1 phosphorylation on tyrosine 14 negatively regulates ECM degradation. (A) Forty‐eight hours following siRNA treatment, caveolin 1‐depleted A375MM cells were transfected with GFP‐tagged rat pseudo‐phosphorylated (Cav 1^Y14D) and non‐phosphorylatable (Cav 1^Y14F) caveolin 1. Lysates from mock (Mock), caveolin 1 siRNA (Cav 1‐KD), caveolin 1^WT (Cav 1^WT), caveolin 1^Y14D (Cav 1^Y14D) and caveolin 1^Y14F (Cav 1^Y14F) transfected cells were subjected to Western blotting and probed with anti‐caveolin antibody to detect endogenous and exogenously expressed caveolins and anti‐GAPDH as a loading control. (B) After the above treatments, cells were plated on gelatin and degradation assays were performed. Caveolin 1^Y14F rescued the normal levels of ECM degradation, whereas caveolin 1^Y14D expression failed to do so. Error bars show mean standard deviations. (C) A375MM cells transfected with caveolin 1^WT, pseudo‐phosphorylated (Cav 1^Y14D), non‐phosphorylatable (Cav 1^Y14F) or non‐transfected (NT) controls were processed for filipin staining, to detect plasma membrane cholesterol levels. Scale bar 10 μm.

To further confirm that caveolin was an essential regulator of cholesterol trafficking in invadopodia biogenesis, we took advantage of two dominant‐negative caveolin 3 N‐truncated mutants [35]. Caveolin 3^DGV lacks the first 53 amino acids while maintaining the cytoplasmic C‐terminus and the upstream scaffolding domain, a membrane‐attachment portion known to bind and regulate several signalling molecules [17]; caveolin 3^DGV has been shown to block cholesterol delivery to the plasma membrane as well as SV40 uptake. Caveolin 3^KSY, instead, lacks the first 107 amino acids and maintains only the cytosolic C‐terminal domain; the expression of caveolin 3^KSY was also reported to block SV40 entry, but at variance with caveolin 3^DGV, did not interfere with cholesterol transport. When A375MM cells were transiently transfected with caveolin 3^DGV, a 45% decrease in ECM degradation was observed as compared to caveolin 3^WT‐transfected cells (Fig. 7A). Total cell cholesterol was not affected (data not shown), whereas plasma membrane levels were found to be considerably decreased as revealed by filipin staining (Fig. 7B). Interestingly, cholesterol re‐addition completely recovered normal invadopodia formation and ECM degradation (Fig. 7A). To exclude that the inhibitory effect of caveolin 3^DGV expression was due to a perturbation of caveolae‐mediated endocytosis, cells were transfected with caveolin 3^KSY. When ECM degradation was quantified, we actually found a slight increase (Fig. 7A). An additional observation ruling out a role for caveolin‐mediated endocytosis at invadopodia, stems from correlative‐light microscopy experiments showing that caveolin 1, although not excluded, is not specifically enriched, nor are caveolae observed, at invadopodia (Fig. 7C–E). This also indicates that non‐caveolar caveolin 1 is involved in invadopodia formation. Of note, the expression of caveolin 3^DGV in caveolin 1‐depleted cells produced an additive effect leading to an 80% decrease in ECM degradation (Fig. 7A).

Figure 7.

Inhibition of cholesterol transport to the plasma membrane impairs ECM degradation at invadopodia. A375MM cells transfected with caveolin 3^WT, caveolin 3^DGV and caveolin 3^KSY were assayed for ECM degradation. (A) A 50% inhibition was found in caveolin 3^DGV‐transfected cells (Cav 3^DGV), whereas a slight increase was observed after caveolin 3^KSY transfection (Cav 3^KSY), compared to caveolin 3^WT (Cav 3^WT). Cholesterol re‐addition (Chol) to caveolin 3^DGV‐expressing cells rescued cells to control degradation levels. When caveolin 3^DGV was expressed in caveolin 1‐depleted cells, degradation was inhibited 80% (Cav 1 KD + 3^DGV). (B) Filipin staining of caveolin 3^DGV‐transfected samples weakly labelled the plasma membrane as compared to non‐transfected cells. Scale bar 10 μm. Correlative light electron microscopy cross‐section of a degrading cell (C) indicates that neither caveolin 1 nor caveolae is enriched in the invadopodial structure (enclosed in the white outline). Asterisk indicates extracellular space (D), Another field is shown to illustrate caveolin 1 staining as black enhanced immunogold speckles (E). Scale bar 10 μm in C and 1 μm in D and E.

Altogether these experiments suggest that invadopodia biogenesis and function rely on correct sorting of cholesterol to the plasma membrane and thus further support the hypothesis that invadopodia feature a high CEMM component.

Role of caveolin 1 phosphorylation in invadopodia biogenesis

Our experiments suggest that caveolin 1 regulates invadopodia biogenesis through the modulation of cholesterol homeostasis and trafficking. The activity of caveolin 1 itself can be regulated by phosphorylation. Specifically, caveolin 1 is phosphorylated by Src‐family kinases (SFK) on tyrosine 14 in the N‐terminal cytosolic domain. This post‐translational modification triggers, upon cell detachment, caveolin 1‐mediated CEMM internalization from the plasma membrane [36]. Also, MCD‐mediated cholesterol depletion increases phosphorylation on Tyr14 (Leyt, 2007); hence there is an inverse correlation between plasma membrane cholesterol and Tyr14 phosphorylated caveolin 1 levels. We thus hypothesized that, if invadopodia formation relies on correct levels of cholesterol, phosphorylated caveolin 1 should clear CEMM from the plasma membrane, and thus inhibit invadopodia formation and ECM degradation. To test this, we first verified cholesterol levels at the plasma membrane in caveolin 1‐depleted cells transfected with either caveolin 1^WT, the pseudo‐phosphorylated caveolin 1 mutant (caveolin 1^Y14D), or with the non‐phosphorylatable mutant (caveolin 1^Y14F), using mock‐transfected cells as a control (Fig. 8A). Filipin staining revealed comparable labelling in mock, as well as in caveolin 1^WT‐ and caveolin 1^Y14F‐expressing cells, whereas a strong reduction was observed consequently to caveolin 1^Y14D transfection (Fig. 8C). When cells were subjected to a degradation assay, caveolin 1^Y14F rescued the phenotype impaired by caveolin 1 knock‐down, whereas caveolin 1^Y14D failed to restore normal levels of invadopodia formation and ECM degradation (Fig. 8B). This would suggest that phosphorylation of caveolin 1 on Y14 is inhibitory towards invadopodia biogenesis, in agreement with the interpretation that invadopodia formation requires the presence of CEMM at the plasma membrane. We cannot exclude that caveolin 1^Y14D could impair invadopodia formation via the inhibition of upstream signalling cascades, in addition to a direct effect on cholesterol homeostasis. Indeed caveolin 1, when phosphorylated on Tyr14, is a docking site for the SH2 domain of C‐terminal Src kinase (Csk), which in turn inactivates Src [37, 38]; this would act as an inhibitory signal as invadopodia formation and functionality require high levels of Src activation [5]. To verify this possibility, we investigated if a correlation between caveolin 1^Y14D‐expression and Src activation levels exists, by staining caveolin 1‐depleted cells expressing caveolin 1^Y14F and caveolin 1^Y14D for phosphorylated, i.e. active, Src. We found that phospho‐Src levels decreased 20% in caveolin 1^Y14D‐expressing cells compared to untransfected cells (Fig. 9A–C and Fig. S2). Strikingly, caveolin 1^Y14F‐transfected cells exhibited a 40% increase of activated Src (Fig. 9D–F and Fig. S2). Our data show that caveolin 1 tyrosine‐phosphorylation inhibits invadopodia biogenesis probably due to removal of cholesterol from the plasma membrane and possibly with a contribution deriving from the inhibition of Src activity. These experiments also validate caveolin 1^Y14D as a bona fide phosphorylated caveolin 1.

Figure 9.

Caveolin 1 phosphorylation on tyrosine 14 decreases active Src levels. A375MM caveolin 1 knock‐down cells were transfected with GFP‐tagged caveolin 1^Y14D (A) and caveolin 1^Y14F (D) and plated on unconjugated gelatin coated coverslips for 16 hrs. Samples were then labelled with an antibody recognizing phosphorylated Src (B and E, respectively). Merged images are also shown (C and F, respectively). Scale bar 10 μm.

Discussion

We show here that invadopodia formation, function and structural integrity depend on appropriate levels of plasma membrane cholesterol, and that invadopodia display the properties of cholesterol‐rich plasma membrane domains. Furthermore, caveolin 1 appears to be a central regulator of invadopodia formation, by controlling cholesterol homeostasis at the plasma membrane.

These conclusions are supported by the fact that cholesterol and sphingolipid depletion impairs invadopodia formation and their persistence. Also, a pool of cortactin and dynamin 2, typical invadopodia components, specifically appeared to associate with the CEMM fraction, when cells displayed prominent invadopodia and were actively engaged in degradation. This, together with the fact that invadopodia exhibited resistance to TX100 extraction, further sup ports the notion that they feature a high CEMM component. The role of cholesterol at invadopodia was also confirmed by the fact that impairment of invadopodia formation and ECM degradation after caveolin 1 knock‐down was reverted by simple provision of cholesterol. In addition, the inhibitory effect of caveolin 3^DGV expression, a mutant known to block cholesterol transport to the plasma membrane, was similarly reverted by provision of cholesterol. As caveolin, nor caveolae, are concentrated at invadopodia, we conclude that caveolin 1 exerts its function in invadopodia formation by regulating cholesterol balance at the plasma membrane.

The specific role played by cholesterol in invadopodia biogenesis and function is currently unknown, although a number of suggestions can be inferred from current knowledge on known cholesterol functions. First, it is a key molecule in maintaining membrane integrity and functionality. Also, by virtue of its peculiar structure, cholesterol promotes the segregation of saturated phospholipids and glycosphingolipids from bulk unsaturated phospholipids. This in turn leads to formation of lipid platforms (the CEMM), where specific proteins, owing to their affinities for these lipids, reside permanently or transiently, with a consequent fine‐tune regulation of downstream signalling activity. Hence one or more of these functions might apply to invadopodia; for instance by contributing to local integrin signalling, to support polarized trafficking of specific components to sites of degradation and/or cytoskeletal remodelling.

In connection with our findings, it should be mentioned that a relationship between caveolin 1 and cancer, although controversial, has recently emerged [39]. First, caveolin 1 has been shown to have an important role in cell migration and invasion, by regulating E‐cadherin function and metalloprotease localization and activity. Caveolin 1 apparently plays a tumour suppressive function by virtue of its caveolin scaffolding domain (residues 82–101), which binds and inhibits several pro‐oncogenic signalling molecules, such as SFK and MAP kinases. On the other hand, caveolin 1 has been defined as a prognostic marker for aggressive prostate, pancreatic and esophageal carcinoma, and an important mediator of hormone‐dependent signal transduction [40, 41, 42]. In prostate cancer, the progression of malignancy has been shown to directly correlate with caveolin 1 expression levels [43], and caveolin 1‐null mice with transgenic prostate carcinoma exhibited a significant reduction of primary tumour mass and metastasis formation [39]. Also, caveolin 1 can be secreted by prostate cancer cells to act as a pro‐angiogenic autocrine/paracrine factor to promote cancer survival and progression [44]. Such apparently conflicting observations might be reconciled by considering that plasma membrane cholesterol homeostasis is so crucial for cell functions that minor imbalances, in one direction or the other, might lead to substantial effects.

Intriguingly, prostate cancer was shown many years ago to be highly enriched in cholesterol (Swyer 1942) and, more recently, to be promoted by cholesterol accumulation at the plasma membrane [34]; furthermore, elevation of circulating cholesterol sustained tumour growth and cell survival in tumour xenografts and in vitro[34]. These notions establish a connection between cholesterol, cancer and caveolin 1.

In general, elevated blood cholesterol is a serious medical problem that is mainly dealt with the administration of statins. These are widely used (up to 10% of the general population) blood cholesterol‐lowering drugs that significantly reduce cardiovascular morbidity and mortality. Population studies suggest a correlation between the administration of statins and the reduction of melanomas, colon, breast and prostate cancer [45] although their anti‐tumoural properties are far from established [46, 47, 48]. In vitro studies showed that statins exert pleiotropic effects on cell survival, angiogenesis, cell migration and proliferation, cytoskeleton remodelling and to affect invasion and metastasis of melanoma cells [49, 50, 51]. These activities have been correlated mainly to the inhibition of protein isoprenylation. In fact, statins inhibit 3‐hydroxy‐3‐methylglutaryl coenzyme A (HMG‐CoA) reductase, the enzyme responsible for the conversion of HMG‐CoA into mevalonate, the obliged step for the progression of the mevalonate pathway leading to both isoprenoid precursor and cholesterol biosynthesis. Although in vitro the HMG‐CoA reductase inhibitors affect the isoprenylation of many proteins, thereby altering their activities, it is likely that statin therapy mainly functions by targeting HMG‐CoA reductase in the liver, rather than in peripheral tissues, thereby potently reducing circulating LDL and hence cholesterol availability in the general circulation [45]. Thus, at standard blood cholesterol‐lowering doses, the anti‐tumoural properties of statins could be related to cholesterol decrease rather than to inhibition of protein isoprenylation. This knowledge provides a further interesting link with our findings.

CEMM act as platforms for the function of, and interaction between, molecules involved in the control of cell signalling, growth, remodelling, trafficking and apoptosis, and a growing body of evidence indicates that altering such microdomains could target different cancer types. Our findings that invadopodia formation/function rely on CEMM integrity and levels at the plasma membrane are in line with current thinking on the relationships between cholesterol, caveolin and cancer. They also reinforce the notion of CEMM as pharmacological targets in agreement with the new concept of ‘membrane‐lipid therapy’[52], according to which protein function can be modulated by altering the surrounding lipid environment, crucial in determining protein localization and activity.

New druggable targets are urgently needed to treat tumours that, as melanomas, exhibit persistent drug‐resistance and are characterized by high incidence of metastasis, morbidity and mor tality rate. Invadopodia are powerful model structures to study ECM degradation, a rate‐limiting step in many invasion and metastatic processes, and might provide new avenues into drug target discovery. In particular, although further work is needed, our results support the idea that the pharmacological alteration of cholesterol and sphingolipid levels or caveolin function could be a possible avenue to modulate invasive potential in alternative to strategies aimed at inhibiting proteolytic activities.

Supporting information

Fig. S1. Effect of cholesterol depletion on pre‐formed invadopodia. A375MM cells were plated on gelatin and left untreated (NT) or incubated with 4 μM lovastatin/0, 25 mM mevalonate (L+M) for 16 hrs. Treated cells were then subjected or not to acute cholesterol extraction with 10 mM methyl‐betacyclodextrin (MCD) for 40 min., incubated for further 3 hrs and then fixed and stained with phalloidin. The percentage of cells with invadopodia over total cells was then calculated, considering at least 30 random fields per each sample.

Fig. S2. Phospho‐Src levels in caveolin 1Y14D‐ and caveolin 1Y14F‐transfected cells. A375MM caveolin 1 knock‐down cells were transfected with GFP‐tagged caveolin 1Y14D and caveolin 1Y14F and plated on unconjugated gelatin coated coverslips for 16 hrs. Samples were then labelled with an antibody recognizing phosphorylated Src and Alexa Fluor 633‐conjugated phalloidin. Levels of phosphorylated Src were calculated as the ratio fluorescence/cell area, considering transfected and untransfected (NT) cells in the same field and expressed as a percentage of pSrc intensity in untransfected cells. The standard deviation was calculated among 20 fields per each sample.

Please note: Blackwell Publishing are not responsible for content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Supporting info item

Supporting info item