EphA2 Induces Metastatic Growth Regulating Amoeboid Motility and Clonogenic Potential in Prostate Carcinoma Cells

Among members of the unique family of receptor tyrosine kinases, EphA2 is frequently upregulated in a variety of cancers and tumor cell lines, including breast, prostate, liver, non–small cell lung, and colon cancers, melanoma, ovarian cancer, and neuroblastoma (1). Unlike other families of receptor tyrosine kinases, which bind to soluble ligands, Eph receptors interact with cell surface–bound ephrin ligands, thus Eph–ephrin interaction stimulates a bidirectional signaling. Despite other classic oncogenes, Eph receptor signaling does not seem to convey a proliferative signal in many cell types, rather it can affect cell survival, cell–cell and cell–matrix attachment modulating tumor cell motility, invasion and metastasis (2). Even if a clear correlation between EphA2 and carcinogenesis has been underlined, the role of EphA2 in in vivo tumor growth and metastasis spreading is still a critical question. First of all, EphA2 expression is upregulated during tumor-associated angiogenesis (3, 4) and, accordingly, soluble EphA2-Fc receptor treatment results in decreased tumor vascular density, tumor volume, and cell proliferation in the 4T1 model of metastatic mammary adenocarcinoma (3, 5). Besides the effect on tumor vasculature, EphA2 retains a role in tumor progression. First, the level of EphA2 expression in human prostate cancer cell lines relates to their metastatic potential in vivo (6) and EphA2 overexpression confers malignant transformation and tumorigenic potential in MCF10A normal epithelial cells (7), and increases invasiveness of pancreatic adenocarcinoma cells (8, 9). Furthermore, overexpression of dominant-negative EphA2 mutant or a kinase inactive form results in decreased tumor volume and increased tumor apoptosis (10, 11). Recently, Nasreen and colleagues (12) showed that silencing EphA2 expression by using short interfering RNA (siRNA) inhibits the proliferation and haptotaxis of malignant mesothelioma cells. In agreement, Landen and colleagues (13) showed that the therapeutic delivery of EphA2 siRNA into an orthotopic mouse model of ovarian cancer reduced tumor growth when compared with a nonsilencing siRNA. Moreover, Brantley-Sieders and colleagues (14) showed that EphA2 has a positive role during mammary tumor onset and growth in the MMTV-Neu transgenic mice model but not in mice overexpressing the polyomavirus middle T antigen, suggesting that, at least in breast carcinoma, EphA2 role in tumor progression depend on oncogene/tumor suppressor context. Recently, Parri and colleagues (15) showed that in prostate carcinoma cells EphA2 stimulation causes the retraction of the cell body and the redirection of cell migration by activating the Src–FAK complex, leading to a Rho-dependent acto/myosin contractility response. Moreover, in a model of in vivo prostate carcinoma metastasis, we observed that the disruption of EphA2 kinase activity strongly inhibits PC3 cell motility and formation of metastatic colonies (11). Furthermore, we reported that PC3 cells expressing EphA2 display an amoeboid motility style, whereas cells expressing kinase-deficient EphA2 mutants lose this particular ability to move without proteolitically destroying extracellular matrix (ECM; refs. 11, 15). Besides, we have shown that in melanoma cells EphA2 reexpression converts the migratory style from mesenchymal to amoeboid-like, conferring them a strong invasive advantage leading to a successful metastic program (16).

In this background, we report here a central role of EphA2 expression for prostate carcinoma and metastatic spreading. Indeed, we show that the silencing of EphA2 is able to eliminate the skill of PC3 cells to move through an amoeboid motility style. This event is associated with a decrease in the stem cell markers, finally leading to an impairment of tumor growth and metastatic dissemination.

Unless otherwise specified, all reagents were obtained from Sigma. Anti-EphA2 antibodies were from Upstate Biotechnology Inc. The invasion chambers were from Corning Costar. The Matrigel matrix, anti-Rac1 antibody, and monoclonal anti-human chemokine (C-X-C motif) receptor 4 (CXCR4) antibody were from RD System. Cell Trace CFSE and Calcein were from Invitrogen. Ilomastat was from Chemicon International. Type I collagen, the Fluorescein isothiocyanate (FITC) mouse anti-human CD44 (clone G44-26), and PE mouse anti-human CD24 (clone ML5) antibodies were from BD Bioscience. The Amplite Universal Fluorimetric MMP Activity Assay Kit was from AAT Bioquest Inc. Anti-mouse Alexa 488 antibody was from Molecular Probes.

Prostate cancer cell lines (PC3) were purchased from ATCC; human umbelical vein endothelial cells (HUVEC) were obtained from BioWhittaker and cultured according to the manufacturer's instruction. PC3 cells were cultured in Ham's F12 medium and maintained in 5% CO₂-humidified atmosphere.

PC3 cells were transfected with empty vector (wtPC3) or 4 different shRNA plasmids (SureSilencing shRNA Plasmids from SuperArray) validated specifically to knock down the expression of EphA2. Cells were selected by G418 treatment and a pool of stably transfected cells was generated to avoid clonal selection. Among the 4 different constructs provided by the kit, 1 caused the almost complete knock down expression of EphA2, as evaluated by Western blot and FACScan analyses. To exclude off-target effects due to the use of a single plasmid, a second construct provided by the same kit has been used to generate an additional EphA2 knockdown pool of stably transfected cells (EphA2-silenced PC3 pool 2, see Supplementary figure S1).

A total of 1 × 10⁶ cells were lysed for 20 minutes on ice in 500 μL of complete radioimmunoprecipitation assay (RIPA) lysis buffer [50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1% NP40, 2 mmol/L EGTA, 1 mmol/L sodium orthovanadate, 1 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin, 10 μg/mL leupeptin]. Lysates were clarified by centrifuging, separated by SDS-PAGE, and transferred onto nitrocellulose. The immunoblots were incubated in 3% bovine serum albumin (BSA), 10 mmol/L Tris-HCl (pH 7.5), 1 mmol/L EDTA, and 0.1% Tween 20 for 1 hour at room temperature and were probed first with specific antibodies and then with secondary antibodies.

PC3 cells were serum starved for 24 hours and then 3×10⁵ cells were seeded onto Matrigel-precoated Boyden chamber (8-mm pore size, 6.5 mm diameter, 12.5 μg Matrigel/filter) with or without 50 μmol/L ilomastat. In the lower chamber, complete medium was added as chemoattractant. Following 24 hours of incubation, the inserts were removed and the noninvading cells on the upper surface were removed with a cotton swab. The filters were then stained using the Diff-Quik kit (BD Biosciences) and photographs of randomly chosen fields are taken.

To examine colony growth in soft agar, 2×10⁴ PC3 cells were plated per 35-mm dish in growth medium supplemented with 0.3% agar. This suspension was layered over 0.9% agar in growth medium. Colony growth was scored after 3 weeks under an inverted microscope.

Serum-free medium from confluent monolayer of cells was collected and 20 μL were added to the sample buffer (SDS 0.4%, 2% glycerol, 10 mmol/L Tris-HCl, pH 6.8, 0.001% bromphenol blue). The sample was run on a 10% SDS gel containing 0.1% gelatin. After electrophoresis, the gel was washed twice with 2.5% Triton X-100 and once with reaction buffer [50 mmol/L Tris-HCl (pH 7.5), 200 mmol/L NaCl, 5 mmol/L CaCl₂). The gel was incubated overnight at 37°C with freshly added reaction buffer and stained with Laemli Comassie blue solution. Areas of gelatinase activity appear as clear bands against a dark background.

Matrix metalloproteinase (MMP) activity was measured with Amplite Universal Fluorimetric MMP Activity Assay Kit according to the manufacturer's instructions. Briefly, serum-free medium from confluent monolayer of cells was collected and 5 μL were added to 4-aminophenylmercuric acetate (AMPA; (1 mmol/L) at 37°C for 1 hour to detect MMP-2 activity and at 37°C for 2 hours to detect MMP-9 activity. A 50 μL portion of the mixture was then added to 50 μL of MMP Red substrate solution. After 60 minutes of incubation the signal was read by fluorescence microplate reader with excitation (Ex)/emission (Em) = 540 nm/590 nm.

Confluent monolayer of cells serum starved for 24 hours were washed twice with PBS and then incubated for 10 minutes at 37°C with 0.25% trypsin. Trypsin was blocked with 0.5 mg/mL soybean trypsin inhibitor. A total of 10⁶ cells were then resuspended in 1 mL of fresh serum-free medium and seeded onto dishes precoated with 10 μg/mL fibronectin.

HUVECs were grown to confluence on the separating filter of a Boyden chamber (8-mm pore size, 6.5 mm diameter). HUVECs were activated with 10 ng/mL TNFα for 90 minutes. Thereafter, culture media were changed for fresh media and cells incubated for an additional 2.5 hours. PC3 cells (3×10⁵), serum starved for 24 hours and treated with Calcein AM, were seeded onto HUVEC cells monolayer. In the lower chamber, complete medium was added as chemoattractant. Following 16 hours of incubation, the insert was removed and the noninvading cells on the upper surface were removed with a cotton swab. The number of cells that have migrated to the lower face of the filter was evaluated by counting the green cells using an inverted fluorescent microscope.

Total RNA from wtPC3 and EphA2-silenced PC3 cells was extracted using RNeasy (Qiagen) according to the manufacturer's instructions. Strands of cDNA were synthesized using a high-capacity cDNA reverse transcription kit (Applied Biosystem) using 1 μg of total RNA. For quantification of CXCR4 mRNA, real-time PCR, using Power SYBR green dye (Applied Biosystem) was done on a 7500 Fast Real Time PCR system (Applied Biosystem). The primers for CXCR4 were 5′-AGCATGACGGACAAGTACAGG-3′ (forward) and 5′-GATGAAGTCGGGAATAGTCAGC-3′ (reverse). Data are normalized to those obtained with glyceraldehyde-3-phosphate deydrogenase primers. Results (mean ± SD) are the mean of 3 different experiments.

Cells were directly lysed in RIPA buffer, the lysates were clarified by centrifugation, and RhoA-GTP or Rac-GTP levels were quantified. Briefly, lysates were incubated with 10 μg of Rhotekin–glutathione S-transferase (GST) fusion protein (Becton Dickinson) or p21-activated kinase–GST fusion protein, both absorbed on glutathione-Sepharose beads for 1 hour at 4°C. GST pulled-down immunoreactive RhoA or Rac1 were then quantified by Western blot analysis. Lysates were normalized for RhoA or Rac1 content by immunoblot.

The percentage of cells undergoing apoptosis after a 24 or 72 hours suspension treatment was assayed by the Annexin-VFLUOS Staining Kit (Roche Applied Sciences) according to the manufacturer's instructions. Briefly, 5×10⁵ PC3 cells were washed with ice-cold PBS and resuspended in 100 μL of binding buffer (HEPES buffered saline solution with 2.5 mmol/L CaCl2). FITC-labeled annexin V and 10 ng/mL propidium iodide (PI) were added. Cells were then incubated for 10 minutes in the dark at room temperature. After the addition of 400 μL of binding buffer and agitation, flow cytometry was done using a BD FACS Canto (BD Biosciences). The analyzer threshold was adjusted on the flow cytometer channel to exclude most of the subcellular debris to reduce the background noise. The totality of annexin V⁺/PI⁻ (early apoptotic) and annexin V⁺/PI⁻ cells (late apoptotic) were considered apoptotic.

To determine the surface expression of CXCR4, EphA2, CD44, and CD24, 10⁶ cells were detached nonenzymatically with 2.5 mmol/L EDTA and incubated with the antibodies according to the manufacturer's instructions in PBS containing 1% BSA for 1 hour at 4°C. After washing with PBS/1% BSA, cells were incubated with Alexa 488–labeled anti-mouse antibodies for 30 minutes at 4°C. On washing, a flow cytometer analysis was done.

Reconstruction by time-lapse video microscopy and confocal microscopy was conducted. Cells were detached by EDTA (2 mmol/L), washed, incorporated into 3-dimensional (3D) collagen/Matrigel lattice (0.55 mg/mL Matrigel and 1.11 mg/mL type I collagen) and monitored by time-lapse video microscopy according to Friedl and colleagues (17). For 3D time-lapse confocal microscopy (Leica-SP5 system), cells within the lattice were labeled by 5 μmol/L Cell Trace CFSE, scanned at 3-minute time intervals for simultaneous fluorescence and back scatter signal (reflection), and reconstructed. 3D motility of cells is shown by time lapse of xyz t analysis (3D analysis during time). Movies are a 2-dimensional (2D) projection (xy) of all stacks during a time course. xy 2D migration of cells has been excluded by analysis of zx axis movements during the same time course. For speed quantification, locomotor parameters of cells incorporated into 3D collagen lattice were obtained by computer-assisted cell tracking and reconstruction of the x and y coordinates of the cell paths. The population speed is the ratio between the total length of the path of single cells and the time.

Cells were detached using Accutase (Sigma). For prostasphere formation, single cells were plated at 150 cells/cm² on low attachment 100-mm plate (Corning) in DMEM/F12 (Invitrogen) supplemented with B27 and N2 (Invitrogen), 5 μg/mL insulin, 20 ng/mL basic fibroblast growth factor, and 20 ng/mL epidermal growth factor. Cells were grown under these conditions for 10 days and then prostaspheres were photographed.

Experiments were done in agreement with national guidelines and approved by the ethical committee of Animal Welfare Office of Italian Work Ministry and conform to the legal mandates and Italian guidelines for the care and maintenance of laboratory animals. Male SCID-bg/bg mice (6- to 8-week old; Charles River Laboratories International, Inc.) were injected subcutaneously (s.c.), both in the right and left lateral flanks, with cells mixed in a 1:1 volume ratio with Matrigel, in a final volume of 200 μL. Animals were monitored, tumor size was measured by a caliper, and tumor volumes determined by the length (L) and the width (W): V = (LW²)/2.

Male CD1 nude mice were purchased from Charles River. Mice were maintained under the guidelines established by our institution (University of L'Aquila, Medical School and Science and Technology School Board Regulations, complying with the Italian Government Regulation n.116, January 27, 1992). The procedure of heart injection of prostate cancer cells in nude mice has been previously described (18). Briefly, 1 × 10⁵ cells in 0.1 mL of saline solution, were injected in the left ventricle of 4-week-old nude mice previously anesthetized with a mixture of ketamine (25 mg/mL)/xylazine (5 mg/mL). The number of mice analyzed was 8 per group. The development of tumor colonies in the whole skeletal apparatus was monitored at times by radiography using a Faxitron cabinet X-ray system (Faxitron X-ray Corp.). All animals were subjected to accurate necroscopy for the evaluation of the presence of tumor colonies in other anatomic sites.

To deeply investigate the role of EphA2 overexpression in prostate cancer and its role in cell motility and tissue invasion, we silenced EphA2 expression by RNA interference. Thus, we generated a pool of PC3 cells in which EphA2 is stably knocked down (EphA2-silenced PC3 cells) with respect to control cells (wtPC3 cells), as indicated by Western blot and flow cytometry analyses (Fig. 1A and B)

The long route of prostate cancer cells to metastasize a distant organ is composed of a series of sequential steps (19, 20), involving entry into the systemic vasculature through the activation of transendothelial migration, sustain of cell survival, homing of cancer cells to the target tissue, adhesion to the new localization and growth of the colony to generate metastases. To understand the importance of EphA2 expression to produce successful metastases, we analyzed EphA2 involvement in each step of the metastatic progression. The activation of transendothelial migration, a step crucial for both extravasation and intravasation of migrating metastatic cells, has been analyzed by a transendothelial motility assay to test the ability of tumor cells to penetrate a monolayer of endothelial cells. As shown in Figure 1C, the efficiency of cells to migrate through an endothelial cell barrier is very similar between EphA2-silenced PC3 cells and wtPC3.

In some cellular models, EphA2 expression represents a way for cells to escape from anoikis as shown in pancreatic adenocarcinoma cells (9), or to avoid caspase-9–mediated apoptosis as evidentiated in malignant mesothelioma cells (12). To test whether in PC3 cells the expression of EphA2 could confer resistance to anoikis, we analyzed cell survival to 24 hours and 72 hours suspension by flow cytometry analysis. As shown in Figure 1D, in this cellular model EphA2 expression is not responsible for any difference in cell survival between EphA2-silenced PC3 cells and wtPC3 cells.

To generate a metastatic tumor, colony prostate carcinoma cells must adhere to the new niches. We therefore investigated the adhesive properties of wtPC3 and EphA2-silenced PC3 cells in a cell adhesion assay on fibronectin-coated plate. As shown in Figure 1E no remarkable differences have been observed between these two populations, suggesting that in this context, EphA2 is not involved in regulating the adhesive properties of prostate cancer cells.

Recent results from our laboratory evidentiated a crucial role of EphA2 in the switch between mesenchymal to amoeboid motility style in cancer cells (16). In particular, we showed that the reexpression of EphA2 in melanoma cells, through a mesenchymal-amoeboid transition (MAT), causes the acquisition of a MMP-independent/RhoA-dependent amoeboid strategy, conferring an increased motility and invasion to cells. Finally, this shift allows melanoma cells to successfully colonize lungs and peritoneal lymph nodes (16). In addition, we already reported that PC3 cells are able to invade through a Rho-dependent amoeboid motility style (11). In this light, we investigated the role of EphA2 in determining the style of prostate cancer cell motility. First, we analyzed the invasive properties of wtPC3 and EphA2-silenced PC3 cells across a Matrigel barrier in the presence of the MMP inhibitor ilomastat. As shown in Figure 2A, ilomastat greatly inhibits EphA2-silenced PC3 cells invasion, whereas wtPC3 motility is only marginally affected by this inhibitor. These data suggest a MMP-dependent motility style for EphA2-silenced PC3 cells, whereas EphA2 expression is associated with a MMP-independent motility style. Accordingly, MMP analysis by gelatine zymography reveals an upregulation of MMP expression in EphA2-silenced PC3 cells (Fig. 2B). Furthermore, a quantitative measure of MMP activity confirmed data obtained by gelatin zymography (Fig. 2C). Mesenchymal and amoeboid motilities are differently regulated by the small GTPases RhoA and Rac1, amoeboid motility being correlated with RhoA activation and Rac1 inhibition and mesenchymal motility to the opposite regulation of these GTPases (21). We therefore analyzed RhoA and Rac1 activation and, as shown in Figure 2D, we observed that EphA2 silencing induces inhibition of RhoA and activation of Rac1, in line with a de novo acquired mesenchymal motility strategy of these cells. In keeping with these data, confocal fluorescence-reflection video microscopy analysis shows a slow-moving spindle-shaped phenotype of EphA2-silenced PC3 cells across collagen I lattice with respect to wtPC3 cells (Fig. 3A). Indeed, in keeping with previous data (22, 23), amoeboid wtPC3 cells move faster with respect to mesenchymal EphA2-silenced PC3 cells (Fig. 3B). In addition, in 3D Matrigel lattice, wtPC3 cells show a round shape morphology typical of amoeboid motility, whereas EphA2-silenced PC3 cells have an elongated, spindle-like shape, characteristic of mesenchymal motility (Fig. 3C).

These findings underline a strong correlation between EphA2 expression and a protease-independent migration strategy, commonly referred as amoeboid motility style. Indeed, the silencing of EphA2 impairs prostate carcinoma cells to invade using their amoeboid motility suggesting a limitation of their metastatic potential.

Recently, in both breast and prostate cancers, epithelial-mesenchymal transition (EMT) has been reported to generate cells with stem-like properties (24) and a correlation between EphA2 expression and stem cell markers has been proposed (25). Tumor cell plasticity, that is, the ability of cells to adapt to environmental changes through activation of ad hoc epigenetic programs, includes EMT and MAT. Both EMT and MAT have reported to grant to cancer cells an increased metastatic potential (26–31). We hypothesized that, besides EMT, MAT may be associated with cancer stemness as well. Indeed, previous evidence from our laboratory showed that EphA2 drives melanoma cells toward a conversion of their mesenchymal style to an amoeboid-like behavior, finally leading to a successful lung and lymph node peritoneal metastasis (16). Several reports have acknowledged the CD44^high/CD24^low ratio as a reliable indicator of stemness in prostate and breast carcinoma cells (32, 33). In this light, we analyzed the expression of these two markers in EphA2-silenced PC3 cells. Flow cytometry analysis reveals that EphA2-silenced cells show a decrease in the number of CD44^high/CD24^low positive cells with respect to wtPC3 cells (Fig. 4C). In addition, the analysis of both tumor cell growth in soft agar and the PC3 clonogenic potential reveals that EphA2 is crucial for tumor cell proliferation. Indeed, wtPC3 cells are able to grow in soft agar, forming large and numerous colonies in anchorage-independent conditions (Fig. 4A and B), whereas EphA2-silenced cells completely lose their independence from anchorage. This assay allows to test cell survival at longer time with respect to analysis of suspension culture, thus revealing differences that have not been previously detected by anoikis quantification. In agreement, only EphA2-expressing PC3 cells are able to form large and tightly packed prostasphere (Fig. 4D), resembling the CD44 positive colonies containing stem cells described by Li and colleagues (34). These clones are defined holoclones and are the only ones able to sustain prostate cancer development and serial tumor transplantation in NOD/SCID mice (34). We then compared the tumor-initiating capacities of wtPC3 and EphA2-silenced PC3 cells when injected s.c. into the flanks of SCID-bg/bg mice. We evidentiated that at lower concentrations (2×10⁴ and 2×10³ cells), EphA2 positively influences both latency and rate of tumor growth (Fig. 4E). These results show that EphA2 level correlates with an enrichment in stem cell markers leading to a faster development of cancer. Histologic examination of lungs showed the presence of spontaneous micrometastases only in mice injected with wtPC3 cells (Fig. 4F), thereby confirming that the expression of EphA2 is essential not only for tumor onset but also for the dissemination of cancer cells.

The cancer stem cell (CSC) hypothesis proposes that tumor progression is sustained by CSCs. Indeed, the ability of cancer cells to generate metastases depends on the dissemination of CSCs that have acquired invasive capabilities. These new properties allow the escape of CSCs from primary tumor mass to migrate toward and colonize distant organs. Prostate cancer has a remarkable tendency to metastasize to bone (35). For metastases to occur, the malignant cells must escape the primary tumor, penetrate and circulate through the bloodstream, and then arrest and proliferate in target tissues. The mechanisms that account for bone homing of prostate carcinoma cells have been in part elucidated and the chemokine receptor CXCR4 has been involved. CXCR4 activation affects prostate cancer cell metastatic behavior by increasing both cell adhesiveness and invasiveness (36–38). To evaluate whether the level of CXCR4 is under the control of EphA2, we measured, in wtPC3 and EphA2-silenced PC3 cells, CXCR4 expression by real-time PCR analysis and its surface level by flow cytometry. EphA2-silenced PC3 cells show a decrease in both CXCR4 mRNA and in the cell surface protein expression with respect to wtPC3 cells (Fig. 5 A and B), thereby sustaining the involvement of EphA2 in CXCR4 expression.

To avoid that the differences observed between EphA2-silenced cells and wtPC3 cells may be off-target effects due to the use of a single shRNA vector to deplete EphA2, we generated an additional pool of EphA2 knockdown cells (EphA2-silenced PC3 pool 2) to validate our results (Supplementary Fig. S1A). We then carried out some of the most representative experiments to exclude that the observed effects may be a unique feature of the pool of cells under consideration. As shown in Supplementary Figure S1, we confirmed the differences between wtPC3 and EphA2-silenced cells for what concerns the quantitative measure of MMPs activity (Supplementary Fig. S1B), the tumor cell growth in soft agar (Supplementary Fig. S1C and D), and the cell surface expression of CXCR4 (Supplementary Fig. 1E). Overall, these results prove that the differences observed between wtPC3 and EphA2-silenced PC3 cells are actually due to the expression of EphA2 and are not a specific effect characteristic of a single pool of cells.

Therefore, to evaluate the role of EphA2 expression in the regulation of in vivo metastatic program in particular toward bone, both wtPC3 and EphA2-silenced PC3 cells were used in a rodent model of bone metastasis assay (18). Eight mice were subjected to intracardiac injection of wtPC3 and EphA2-silenced PC3 cells as already described (11). Fifty-two days after heart injection (end point) mice were sacrificed and subjected to a digital scan of total body radiography and to an accurate necroscopic analysis to confirm the presence of both osteolytic and visceral metastases. It is worth noticing that silencing of EphA2 kinase reduces the ability of PC3 carcinoma cells to produce osteolytic metastases up to 50% (Fig. 5C, left). For what visceral metastases are concerned, the effect of silencing EphA2 expression is even more striking, as indicated by the complete abrogation of visceral metastases outgrowth in EphA2-silenced PC3 cells (Fig. 5C, right). We also analyzed the timing of bone metastases outgrowth by a time course radiographic assay, before the experiment end point. The onset of bone metastases in mice injected with EphA2-silenced PC3 cells is strongly delayed with respect to control cells. Indeed, whereas wtPC3 cells developed osteolytic lesions at 36 days (mean value) after intracardiac injection, EphA2-silenced PC3 cells begin the bone metastatic process significantly later (44 days; Fig. 5D and E). Altogether, these data enforce the idea that EphA2 kinase expression is a key determinant for the generation of cells with stem-cell properties endowed with ability to target secondary organs, such as lungs and bone.

Data presented herein show a key role of EphA2 in determining a productive metastatic program mainly dependent on the activation of an amoeboid motility style and the maintenance of stemness. In particular, we disclose that EphA2 is crucial for (i) regulation of an amoeboid-based successful invasive strategy, (ii) increase of stem cell markers and regulation of tumor onset, and (iii) dissemination to bone and growth of metastatic tumors.

Several papers clearly show that EphA2 is able to induce tumorigenesis such as in mammary epithelial cells (7), and in prostate, ovarian, and breast cancer cells (6, 10, 39). The ability of EphA2 to induce tumorigenesis and metastatic dissemination has been correlated with the increased skill of EphA2-overexpressing cells to grow in soft agar, to invade into Matrigel (7) and to increase the resistance to anoikis (9, 12). Among the previously mentioned characteristics concurring to determine a more aggressive phenotype, we would like to introduce 2 new features: the acquisition of an amoeboid motility style due to EphA2 overexpression and the increase in clonogenic potential. Indeed, EphA2 has a crucial role as motility and adhesive factor mainly affecting the balance of Rho, Rac, Cdc42 activities and protein involved in integrin signaling (2). Recently, it has been shown that cancer cells are able to achieve different strategies to evade the primary tumor site and metastasize to distant organs (40). For the invasive migration of cancer cells, at least 2 devices are currently involved: (i) the mesenchymal style depending on extracellular proteolysis and (ii) the proteolysis-independent amoeboid mode depending on the activity of Rho and the Rho-associated coiled-coil–forming protein kinase (ROCK). Indeed, Rho-GTPase is frequently overexpressed in several cancers, whereby increased activity correlates with tumor progression and underlines the different modes of tumor cell motility during invasion and metastasis. The skill of tumor cells to switch between different modes of motility has been shown to limit the efficiency of agents aimed to reduce invasion (41). Recently, besides the well-known importance of mesenchymal motility in cancer cells invasion, growing evidence show the importance of proteolysis-independent movements in neoplastic cells (27, 42, 43). Torka and colleagues (44) showed that human mammary tumor cells displayed an increased activity of ROCK and its downstream effectors, leading to an invasive strategy based on a ROCK-dependent amoeboid motility model. Furthermore, mesenchymally migrating tumor cells, such as HT1080 fibrosarcoma and MDA-MB-231 mammary carcinoma cells, arrest their proteolytic-dependent migration after addition of protease inhibitors and switch to amoeboid behavior, involving vigorous shape change and the ability to squeeze through narrow regions of ECM, thereby rescuing their migration independently from pericellular proteolysis (45). Similarly, Sahai and colleagues (46) showed that ROCK activation causes a transition from mesenchymal to amoeboid movement in HT1080 leading to cortical actin polymerization and cell rounding; finally, this activation of Rho/ROCK signaling allows HT1080 to penetrate thick 3D Matrigel layers. In our cellular model, we observed that EphA2 is a key molecule able to drive prostate carcinoma cells toward an amoeboid motility style. This aptitude, together with the acquisition of an increased clonogenic potential, seems to be crucial for a successful colonization of bone, lung, and visceral metastatic sites. Indeed, EphA2-silenced cells, although showing an increased production of MMPs and invasion through Matrigel by mesenchymal motility, are completely unable to produce lung and visceral metastases and their ability to colonize bones is greatly reduced. Maybe, in the microenvironmental condition that PC3 cells find on their way to give in vivo metastases, the ability to move through an amoeboid motility is more useful rather than the ability to degrade the ECM. In addition, the increase in bone colonization evidentiated in mice injected with wtPC3 cells is in agreement with our data showing that EphA2 exerts a positive control on CXCR4 expression. CXCR4 is fundamental for the acceleration of the metastatic process of prostate tumors (36) and it is a marker of poor cancer-specific survival (47). We may therefore assume that the EphA2 is involved, through the enhancement of CXCR4 expression, in the development of prometastatic signals. In conclusion, we believe that, in keeping with data on melanoma and sarcoma, in prostate cancers a RhoA-dependent, MMP-independent motility style is crucial to successfully invade distant organs. Indeed, we evidentiated a more motile phenotype of wtPC3 cells with respect to EphA2-silenced PC3 cells, as shown by speed measurement. This aptitude, typical of amoeboid movements, allows cells to squeeze between ECM gaps and concur to achieve a successful migratory strategy. In keeping with these data we have already disclosed that EphA2 in PC3 cells leads to a RhoA-dependent actomyosin contractility dependent on Src/FAK activation, suggesting that the FAK/RhoA signaling pathway is mainly responsible for the EphA2-dependent increased motility (15). Accordingly, disruption of EphA2 activation by kinase-deficient EphA2 mutants leads to abrogation of ephrinA1-induced cell rounding, inhibition of FAK-mediated motility response, and inhibition of EphA2-mediated invasion through a Rho-dependent and MMP-independent mechanism (11, 48). Besides, Parri and colleagues (16) have recently shown that EphA2 reexpression in murine melanoma cells, which use a mesenchymal motility, converts this motility style into the amoeboid one, conferring an invasive advantage to lung and lymph node metastases. Hence, the data here presented enlarge the emerging idea of EphA2 has an amoeboid promoting factor to prostate carcinoma, able to ensure an invasive gain adapting cells to environmental changes.

Together with a positive role of EphA2 in warranting an amoeboid motility style, we highlighted a function of EphA2 in the induction of clonogenic potential. Indeed, the CSC hypothesis proposes that tumor growth is sustained by CSCs able to self-renew, giving rise to differentiated progeny and able to reconstitute the whole tumor. CSCs have been characterized in several solid tumors and among these also in prostate cancers (49). Recently, it has been shown that EMT is able to generate cells with CSC properties in breast and prostate cancers and in nontransformed cells (24). Both EMT and MAT are examples of adaptation reactions, which can modify the cell's shape, pattern, and migration mechanism. EMT is a process, which involves the loss of cell–cell junctions and increased cell motility, finally leading to the acquisition of mesenchymal features. Conversely, MAT consists mainly of the abrogation of pericellular proteolysis and strengthening of Rho/ROCK signal pathways, allowing cells to move fast among ECM gaps without destroying it. Tumor cells can switch between different motility styles, thereby adapting migration to the context, finally facilitating a sustained dissemination of single cells. These phenomena are defined “plasticity” and results in migratory ”escape” strategies to allow cancer cell dissemination in tissues. Both EMT and MAT are considered as key features for tumor progression (27, 30, 42–45).

We now propose that the induction of stemness is a general phenomenon associated with shifts in motility styles and therefore, besides EMT, also sustained by MAT. We observed that EphA2 silencing reduces both anchorage-independent cell growth and the expression of stem cell markers. Indeed, we observed that wtPC3 cells, with respect to EphA2-silenced PC3 cells, show (i) higher CD44^high/CD24^low-ratio, (ii) greater capacity to form round colonies with tightly packed cells, (iii) acceleration of the onset of primary tumors (iv) capacity to form spontaneous metastases from s.c. primary tumors, and (v) higher bone osteolytic and visceral metastases. Our data show a small but significant increase of CD44^high/CD24^low ratio in agreement with evidence reporting a dramatic increase in tumorigenicity of CD44⁺ with respect to CD44⁻ prostate cancer cells (50). Thus, the colonies formed by wtPC3 cells resemble the holoclones observed by Li and colleagues (34), who showed that only holoclones expressing high levels of stem and progenitor cell markers contain self-renewing cancer cells that can sustain long-term propagation in culture and tumor development. Altogether these data suggest that EphA2 is crucial for the onset and growth of tumor colonies, and for metastases to occur, with a clear correlation with an increase in the clonogenic potential. Indeed, the decrease in stem cell markers and in cell plasticity observed in EphA2-silenced cells severely impairs the ability of prostate cancer cells to generate distant metastases.

We believe that both the ability of EphA2 to instruct cells to move by amoeboid motility and to generate cells endowed with self-renewal properties, are key features of its function in cancer progression. In conclusion, these data, on one side, support the idea that MAT and clonogenic potential are interrelated phenomena and, on the other side, reinforce the hypothesis that cell plasticity and the achievement of stem-cell properties are crucial for the generation of prostate carcinoma metastases and thus for tumor spreading. EphA2-expressing cells gain an invasive advantage from their amoeboid motility style thus escaping from the primary tumor and then by the concomitant acquisition of clonogenic potential, can successfully develop metastases.

No potential conflicts of interest were disclosed.

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.