Membrane Microvesicles as Actors in the Establishment of a Favorable Prostatic Tumoral Niche: A Role for Activated Fibroblasts and CX3CL1-CX3CR1 Axis

After cellular stimulation in the broad sense of the term, including initiation of a death program by apoptosis, the asymmetric distribution of membrane phospholipids is no longer maintained and the cell releases submicrometric membrane fragments (0.1–1 μm), termed microvesicles (MV), into its environment. MV can be detected in the peripheral blood of mammals. They are characterized by the presence of procoagulant phospholipids (phosphatidylserine, PS) at their surface, which confer a possible prothrombotic potential, and antigens characteristic of their cellular origin (1). It has been shown that the quantity of generated MV is directly proportional to the degree of cellular activation including apoptosis in vitro, and that it is a quasi-universal feature of eukaryotic cells (2, 3). MV contain cytoplasmic components and genetic material. In cardiovascular disorders, numerous studies have led to consider MV genuine pathogenic markers (4), able to remotely initiate or amplify a deleterious process, depending on the cellular type and stimulus at their origin determining their composition. More generally, MV may be considered multifunctional bioeffectors in intercellular communication, and their roles have been investigated in hemostasis, inflammation, immunity, and angiogenesis (1), but only scarce information is available in cancer.

It has to be emphasized that MV or tumor-derived MV (TMV) considered in this study are different from exosomes, another category of smaller vesicles resulting from the endosomal cycle, able to elicit biological responses in various pathophysiologic circumstances, including cancer (5). Exosomes have a more homogeneous protein composition than MV or TMV and do not generally express PS and are viewed as antigen-presenting vehicles (5, 6).

Fibroblasts are associated with tumor cells at all stages of cancer development. It has been proposed they play a central part in promoting migration and invasion of cancer cells, when activated (7). Fibroblasts are thought instrumental in creating a niche for cancer cells, and the activated phenotype has been associated to cancer progression and metastasis, probably sustained through reciprocal signals. Fibroblasts have the capacity to proliferate and/or are kept under an activated state that can support or initiate epithelial hyperplasia and eventual tumorigenesis (8). However, the mechanism leading to fibroblast activation is not yet elucidated, as well as their structural and functional contributions to cancer progression.

CX3CL1/fractalkine is constitutively expressed at low levels in epithelial cells, and is up-regulated in activated epithelia and endothelia (9–11). CX3CL1 exists as both a membrane-bound form promoting firm cell-cell adhesion and a soluble form acting as a chemoattractant for cells expressing its cognate receptor CX3CR1 (9). The expression of CX3CR1 was detected in human prostate cancer cells and the association of CX3CR1-CX3CL1 chemokine members regulates adhesion, migration, and survival of the latter (12).

Several studies have reported a role for MV in the communication between cancer cells and their environment, but data concerning the effect of MV in cancer-stromal cell interactions are missing. Because MV are present in body fluids and within tumor microenvironment, cells could “interpret” MV variations to mutually balance their behavior. In this context, we propose a model where not only cancer cells influence the fate of stromal cells via TMV, but in turn, cancer cells are influenced by MV shed from stromal cells exposed to TMV. Two prostatic carcinoma cell lines PC3 and LnCaP were selected to produce TMV, with respect to their differential metastatic potential, high and low, respectively. Because fibroblasts are major components of the tumor stroma, healthy fibroblasts were used as target cells. The CX3CR1-CX3CL1 axis was considered in these TMV-fibroblast and fibroblasts MV-tumor cell exchanges of information. The results provide a new insight on the crosstalk between tumor and stromal cells mediated by MV, which could be part of the mechanisms governing tumor development.

Fetal bovine serum (FBS) was from Invitrogen. Other cell culture reagents were from Cambrex, actinomycin D (Act.D), staurosporine (St), etoposide (VP-16), 4′,6-diamidino-2-phenylindole (DAPI), propidium iodide (PI), PKH-26 and antibody (Ab) to β-actin, neutralizing goat Ab anti-CX3CL1, and control goat IgG were from Sigma. z-Val-Ala-Asp.fluoromethyl ketone (z-VAD.fmk) and U1026 ([1,4-diamino-2,3-dicyano-1,4-bis(2-aminophenylthio)butadiene]) were from Calbiochem. Ripa Lysis Buffer was from Upstate Biotech.

Monoclonal Abs to matrix metalloproteinase (MMP)2, MMP3, MMP7, MMP9, and polyclonal Abs to MMP13, MMP14 were from LabVisionCorporation. Polyclonal Ab to phosphorylated ERK1/2, Ab anti-EMMPRIN, Ab anti-CX3CR1, and donkey peroxidase-conjugated anti-goat IgG were from SantaCruz, polyclonal Ab against ERK1/2 from CellSignaling, goat peroxidase–conjugated anti-rabbit IgG purchased from Bio-Rad, sheep peroxidase–conjugated anti-mouse IgG from AmershamBiosciences, and streptavidine-FITC conjugate from Rockland. Monoclonal Ab to prostate-specific membrane antigen (PSMA) was from Invitrogen, and polyclonal Ab anti-CX3CL1 was from TorreyPines Biolabs. Neutralizing Ab anti-CX3CR1 was from AbnovaCorporation, and control mouse IgG was from SouthernBiotech. Human blood coagulation factors, chromogenic substrate pNAPEP (Biopep), Annexin-V (AV), and AV-biotin were the same as those previously used (13).

Human prostate adenocarcinoma cell lines PC3 and LnCaP were obtained from the American Type Culture Collection. PC3 cells were cultured in DMEM supplemented with 10% (v/v) heat-inactivated FBS, 100 μg/mL streptomycin and 100 U/mL penicillin, and LnCaP in RPMI 1640 supplemented with 10% heat-inactivated FBS, l-glutamine (2 mmol/L), Hepes (10 mmol/L), sodium pyruvate (1 mmol/L), 1% (v/v) nonessential amino acids, 100 μg/mL streptomycin, and 100 U/mL penicillin. Normal human fibroblasts were a kind gift of Dr. J. Ceraline (Université Louis Pasteur, Strasbourg, France) and cultured in DMEM supplemented with 10% heat-inactivated FBS and gentamicin (20 μg/mL). All cell types were incubated in a humidified 5% CO₂ atmosphere at 37°C. Cell viability and agent cytotoxicity was checked by trypan blue exclusion. All experiments were carried out at 80% to 90% cell confluence. PC3 and LnCaP were incubated without or with Act.D (0.5 μg/mL), St (0.33 nmol/L), and VP-16 (50 μmol/L) to produce TMV (see below). Normal human fibroblasts were cultured without or with 10, 30, or 50 μg/mL protein of different TMV preparations, isolated from nontreated and activated PC3 and LnCaP cells. Apoptotic fibroblasts were evaluated after 48 h of incubation with TMV (1.8% in control fibroblasts, 3% and 3.5% in fibroblasts with 30 or 50 μg/mL of TMV, respectively).

PC3 and LnCaP cells were used for TMV production. Briefly, cells at 80% to 90% confluence were treated with apoptogenic reagents for 24 h (14). Supernatants from both cell cultures were obtained by centrifugation at 750 g for 5 min and then at 1,500 g for 15 min to remove cells and large debris, respectively. TMV from supernatants were pelleted and washed by 3 centrifugation steps (45 min at 14,000 g, room temperature) and recovered in 500 μL HBSS (2). MV from conditioned medium (CM) were obtained from cultured untreated or TMV-activated fibroblasts. After incubation without or with TMV for 48 h, supernatants from fibroblast cultures were processed as described above. Double determination of MV was carried out by measuring their PS content by a functional prothrombinase assay (2) and protein content by Bradford's method (Bio-Rad) with bovine serum albumin (Sigma) as standard. TMV and fibroblast-derived MV isolated from at least five independent preparations were analyzed for each experimental condition. TMV_{PC3 Act.D} PS content was verified to be stable during 48 h of incubation in culture medium.

It has to be indicated that because no ultracentrifugation procedure (100,000 g or higher speed) has been used, exosomes are not expected to be present in MV or TMV preparations (6).

Biotinylated anti-PSMA Ab was insolubilized onto streptavidin-coated microtitration plates and incubated with TMV from PC3 and LnCaP cells. Captured PS⁺ MV (PS⁺-MV) were quantified by prothrombinase assay (2).

To examine the kinetics of interaction between TMV and fibroblasts, the latter were incubated with labeled TMV from Act.D-treated PC3, (TMV_{PC3 Act.D}). Briefly, TMV_{PC3 Act.D} (50 μg/mL protein) were previously labeled with the phospholipid dye, PKH-26 (1 μmol/L), for 10 min at room temperature in the dark, and extensively washed in HBSS/FBS 1:1 by 2 centrifugation steps (45 min at 14,000 g). The TMV_{PC3 Act.D} labeling was stable during 48 h of incubation in culture medium. Fibroblasts were incubated with labeled TMV_{PC3 Act.D}, and at indicated time points, culture medium containing unbound TMV_{PC3 Act.D} was removed from fibroblasts growing in monolayers. Adherent cells were detached and centrifuged. Cells were resuspended in 300 μL of HBSS for flow cytometry analysis. Unbound labeled-TMV_{PC3 Act.D} were evaluated by flow cytometry using a FACScan flow cytometer (Becton Dickinson), and analyses were conducted using CELLQuest software. For fluorescence microscopy, cells were cultured in monolayers, incubated with labeled TMV_{PC3 Act.D} for different incubation times at 37°C, then washed and stained with DAPI. Images (magnification, ×400) were captured with a videographic system and acquired by using Spot software from Diagnostic Instruments, Inc.

After incubation with TMV as described above, fibroblasts were washed thrice with HBSS to eliminate unbound-TMV, and lysed in Ripa Lysis Buffer. Cell lysates were clarified by centrifugation (13,000 g × 3 min, 4°C), and protein content was determined with Bradford protein assay. For detection of TMV-associated metalloproteinases and EMMPRIN, TMV were resuspended in HBSS and samples containing 20 μg protein were separated by 10% SDS-PAGE. Separated proteins were then blotted onto Hybond-enhanced chemiluminescence nitrocellulose membrane (AmershamBiosciences). Blots were probed with specific Ab and developed with the corresponding horseradish peroxidase–conjugated secondary Ab. Proteins were visualized by enhanced chemiluminescence reagents and exposed to CL-XPosure films (Pierce); β-actin staining was used as control. Densitometry analyses were conducted using QuantityOne software (Bio-Rad).

After incubation with TMV as described before, fibroblasts were lysed in lysis buffer. Samples were separated by 10% SDS-PAGE containing 0.2% gelatin B (Sigma) under nonreducing conditions. Gels were washed in 2.5% Triton X-100 for 1 h at room temperature followed by incubation in reaction buffer [50 mmol/L Tris-HCl (pH 7.5), 200 mmol/L NaCl, 5 mmol/L CaCl₂, and 1 μmol/L ZnCl₂] for 40 h at 37°C. Gels were stained with 0.25% Coomassie blue (Sigma) and scanned for determination of gelatinolytic activity.

Fibroblasts were treated with TMV (50 μg/mL protein) and apoptosis was induced simultaneously by Act.D (0.5 μg/mL) in the presence or absence of z-VAD.fmk (50 μmol/L) or U0126 (10 μmol/L) for 24 h. Culture medium was removed from fibroblasts growing in monolayers. Adherent cells were trypsinized, detached, combined with floating cells from the original culture medium, and centrifuged. Cells were resuspended in HBSS containing 1 mmol/L CaCl₂, and incubated for 10 min with AV-biotin (4 μg/mL) and PI (6 μg/mL) at room temperature. Cells were then stained for 5 min with streptavidin-FITC (3 μg/mL), at room temperature in the dark. FITC and PI fluorescence were recorded by flow cytometry. FITC fluorescence was plotted as FL-1 versus PI fluorescence as FL-2, and both were reported in fluorescence intensity arbitrary units under the same settings as described in “Membrane interaction assay.”

Chemotaxis of fibroblasts. The chemotactic motility of fibroblasts was estimated using 24-well Transwell plates with polyethylene-terephthalate filters (8-μm pore size). The lower wells of Boyden chambers (Becton Dickinson) were loaded with DMEM supplemented with 10% FBS as chemotactic factor. After preincubation of fibroblasts with TMV or without (control) for 30 min, they were harvested and resuspended in DMEM containing 0.1% FBS, and loaded at 1 × 10⁵ cells/mL into each of the upper wells. The chamber was incubated at 37°C for 24 h. After incubation, cells were fixed with methanol and stained with crystal violet (1%, w/v). Nonmigrating cells were removed from the upper surface of the filter, and chemotaxis was visualized by microscopy. Images were captured with a videographic system and acquired by using Spot software. Chemotaxis was quantified by counting cells that migrated to the lower side on 4 random fields of the filter at low magnification (×200). Data are expressed as mean ± SE of at least three separate experiments for each condition.

Chemotaxis of cancer cells. To estimate the migration capacity of cancer cells toward fibroblast-CM (CM-Fb), fibroblasts were incubated without (control) or with TMV_{PC3 Act.D} at 37°C for 48 h. CM-Fb were collected and placed in the lower wells with or without Ab to CX3CL1 or irrelevant Ab (8 μg/mL), and used as chemoattractant for cancer cells. PC3 and LnCaP cells were trypsinized and loaded in the upper well of Boyden chambers with or without Ab to CX3CR1 or irrelevant Ab (25 μg/mL), and allowed to migrate for 24 h at 37°C. After incubation, samples were analyzed as described above.

To evaluate the ability of PC3 cells to cross a basement membrane, the Matrigel chemoinvasion assay (Becton Dickinson) was used. The lower chambers were loaded with culture medium from TMV-activated fibroblasts (CM-Fb TMV) or CM from untreated fibroblasts (CM-Fb NT) used as control. CM-Fb was prepared as described above. Cancer cells were loaded into the upper compartment at 3 × 10⁵ cells/chamber and incubated at 37°C for 48 h. After incubation, samples were analyzed as detailed above.

Results were expressed as mean ± SE of at least three separate experiments performed at different culture stages and/or passages, with no obvious difference attributable to the culture conditions. Paired Student's t test was used for statistical analysis. A P value of <0.05 was considered significant.

Kinetics of interaction between tumor cell–derived MVs and fibroblasts. Incubation of fluorescent TMV_{PC3 Act.D} (labeled by PKH-26) with fibroblasts resulted in time-dependent transfer of fluorescence to fibroblasts (Fig. 1A). The increase of fluorescence intensity observed in fibroblasts was accompanied by a decrease of fluorescence intensity of labeled TMV_{PC3 Act.D} counterpart, according to the incubation time. After 48 h at 37°C, most of TMV_{PC3 Act.D} fluorescence disappeared at the benefit of fibroblasts, suggesting transfer from TMV_{PC3 Act.D} to fibroblasts. This was confirmed by fluorescence microscopy analysis (Fig. 1B), in which nuclei of fibroblasts were labeled with DAPI (blue fluorescence) and TMV_{PC3 Act.D} with PKH-26 (red fluorescence). The time-dependent interaction of TMV_{PC3 Act.D} and fibroblasts resulted in MV release from fibroblasts evidenced by increased MV counts occurring in the culture medium measured by flow cytometry, and by a method of MV capture onto AV (2) to detect the presence of PS through its procoagulant potential. As shown in Fig. 1C, there was an inverse relationship between the number of MV and their PS content. The decrease of PS proportion can be explained by the limited PS content of a given cell type (15). In the absence of TMV_{PC3 Act.D}, the basal release of fibroblast MV remained stable and close to the detection limit of the assay (∼1.5 nmol/L PS Eq) during 48 h. Hence, TMV treatment is indeed necessary to elicit MV release from fibroblasts.

Characterization of tumor cell–derived MVs. MMP are endopeptidases that play pivotal roles in promoting tumor progression, including angiogenesis, growth, local invasion, and subsequent distal metastasis (16). The MMP enrichment was evaluated in TMV shed from PC3 and LnCaP cell lines, untreated (control) and activated by different agonists. For MMP-3, only a very weak signal could be detected in all different preparations of TMV (Fig. 2A). MMP-9 and MMP-14 were strongly expressed in TMV_{PC3 Act.D} (Fig. 2B and C). This was the case for MMP-13 in TMV from unstimulated PC3 cells (NT; TMV_{PC3 NT}) and LnCaP cells treated with Act.D (TMV_{LnCaP Act.D}; Fig. 2D). For this reason, only the three different preparations of TMV with strong expression of MMP were considered for assessment of effects on fibroblasts. MMP-7 and MMP-2 were not detected in TMV (data not shown). Because the MMP inducer EMMPRIN has been evidenced at the surface of TMV stemming from tumor cells (17), its presence was verified at the surface of TMV from PC3 and LnCaP cells. As shown, in Fig. 2E, EMMPRIN is mainly expressed in all TMV (TMV_{PC3 NT}, TMV_{PC3 Act.D}, and TMV_{LnCaP Act.D}). In addition, PSMA expressed by normal and neoplastic epithelial cells negatively regulates migration and invasion (18). Using a method of MV antigenic capture and quantification (2), PSMA could be detected mainly in TMV_{LnCaP Act.D} (Fig. 2F,-a), which was confirmed by Western blot (Fig. 2F -b). Hence, TMV present differential expression patterns of MMP depending on the conditions of stimulation. All of them harbor procoagulant PS and tissue factor (TF) activity at their surface (data not shown), and TMV_{LnCaP Act.D} display a stronger expression of PSMA.

Phosphorylation of extracellular signal-regulated kinase 1/2, and active MMP-9 overexpression. Cell migration is critical for many physiologic processes and is often dysregulated in developmental disorders and pathologic conditions, including cancer. Constitutive activation of the extracellular signal-regulated kinase (ERK) and/or phosphatidylinositol-3 kinase/Akt signaling pathways is associated with neoplastic transformation of a variety of human tumor cells (19). The phosphorylation of ERK1/2 was assessed in untreated fibroblasts (NT) or treated with 10, 30, and 50 μg/mL of TMV. As illustrated in Fig. 3A, TMV_{PC3 NT} and TMV_{PC3 Act.D} induced an enhancement of ERK1/2 phosphorylation, this effect was dose-dependent, with a ∼30% increase at 50 μg/mL TMV compared with control (NT). The effect of TMV on the expression of MMP in fibroblasts has been evaluated. Expression of MMP-9 (Fig. 3B) was higher in fibroblasts incubated with TMV (10, 30, and 50 μg/mL) for 24 hours, in a dose-dependent manner. Furthermore, MMP-9 was found active by zymography assay, as evidenced by the gelatinolytic activity detected in fibroblasts exposed to TMV (10, 30, and 50 μg/mL; Fig. 3C). These results indicate that TMV are able to activate signaling pathways in fibroblasts.

TMV promote migration of fibroblasts. TMV induce ERK1/2 phosphorylation and increase active MMP-9 expression in fibroblasts, knowing that these actors are involved in tumor progression. In this line, the capacity of fibroblasts to migrate in the presence of TMV has been assessed. Figure 3D shows that after 24 hours, fibroblasts incubated with 30 and 50 μg/mL of TMV_{PC3 NT} or TMV_{PC3 Act.D}, exhibited a 2- and 3-fold increase of migration, respectively. When fibroblasts were incubated in the presence of 30 and 50 μg/mL TMV_{LnCaP Act.D}, the migration was enhanced by only 1.2- to 1.6-fold, respectively, probably related to the lower metastatic potential of LnCaP cells. However, the apoptotic origin of TMV did not seem to have any influence.

TMV decrease chemosensitivity of fibroblasts. To assess the transfer of TMV-mediated chemoresistance to fibroblasts, the latter were incubated with 50 μg/mL TMV_{PC3 Act.D}, and induced into apoptosis with Act.D (0.5 μmol/L, during 24 hours). Apoptosis ratios in the absence or presence of TMV were measured by double labeling of nonpermeabilized cells with AV and PI. Figure 4A and B show that Act.D-induced apoptosis of fibroblasts (12% of AV⁺/PI⁻ events, reflecting the apoptotic population), was significantly reduced to 7.2% (AV⁺/PI⁻ events) in the presence of TMV. TMV therefore protected fibroblasts from apoptosis by ∼40%, probably through the promotion of partial chemoresistance to Act.D. To investigate the implication of caspases in Act.D-induced apoptosis, the pan-caspase inhibitor, z-VAD.fmk, was used. As reported in Fig. 4B, this inhibitor reduced the degree of apoptosis evoked by Act.D. At 50 μmol/L, z-VAD.fmk significantly inhibited apoptosis, by ∼52%. z-VAD.fmk also prevented apoptosis in fibroblasts treated with TMV, by ∼50%. As expected, these results confirm that caspases play a critical role in apoptosis induced by Act.D in fibroblasts but without a clear differential effect due to TMV.

Because the ERK pathway is implicated in many cellular processes, including apoptosis (20), and ERK1/2 is phosphorylated in fibroblasts after TMV treatment, it was examined whether chemoresistance induced by TMV in apoptotic fibroblasts was mediated by the activation of the ERK pathway. For this, an inhibitor of mitogen activated protein kinase, U0126 (21), was used. As shown in Fig. 4C,-a, this inhibitor promoted an increase (∼25%) of apoptosis evoked by Act.D in fibroblasts, and abolished the chemoresistance induced by TMV in apoptotic fibroblasts. These observations were confirmed by Western blot analysis indicating that ERK1/2 phosphorylation seems strictly correlated with apoptosis chemoresistance induced by TMV (Fig. 4C -b). This suggests that the ERK pathway could be involved in the partial chemoresistance elicited by TMV in Act.D-challenged fibroblasts.

Migration and invasion of PC3 cells toward the CM of stimulated fibroblasts. The above results provide evidence of the effect of TMV on the acquisition of an “activated phenotype” by treated fibroblasts. To further evaluate the consequences of the crosstalk between tumor and stromal cells mediated by TMV, the capacity of PC3 cell migration toward CM from untreated (CM-Fb NT) or TMV-stimulated fibroblasts (CM-Fb TMV) was examined. Basal migration of PC3 cells was observed in the presence of CM-Fb NT, this condition was taken as the control level of migration of PC3 cells (Fig. 5A). CM of fibroblasts incubated for 48 hours with TMV (30 and 50 μg/mL TMV_{PC3 Act.D}), promoted significant migration of PC3 cells. This dose-dependent effect was increased 5- and 7-fold by CM of fibroblasts treated with 30 and 50 μg/mL TMV, respectively (Fig. 5A,-b and A-c), with respect to control (Fig. 5A,-a). Microscopic analysis of membranes from Boyden chambers was performed (Fig. 5B). TMV from LnCaP were also able to promote MV release from fibroblasts that stimulate PC3 migration as observed for PC3 TMV, but to a lower extent, with only ∼40% of the effect of PC3 TMV, in possible relation with the lower metastatic potential of LnCaP cells (data not shown). The next question was is this effect on PC3 migration attributable to soluble factors secreted by fibroblasts in response to TMV treatment, or to MV released by fibroblasts? To address it, the different CM-Fb were centrifuged (see Materials and Methods) to pellet MV, and the corresponding supernatants were examined. These MV-free CM were noted CM without MV. They were incubated with PC3 cells in Boyden chambers for 24 hours and no increase of PC3 migration was observed (Fig. 5A). In addition, to exclude possible contamination by residual TMV (from Act.D-treated PC3 cells) in the CM-Fb TMV, migratory capacity of PC3 cells toward TMV_{PC3 Act.D} was controlled. Under this condition, PC3 cells did not migrate (Fig. 5A,-d and B-d), reinforcing the fact that fibroblasts actively released MV after TMV uptake. Because neither fibroblast MV nor CM on their own were able to induce tumor cell migration that occurred only when together, it can be concluded to a cooperative effect between the two counterparts. Furthermore, as shown in Fig. 5C, CM-Fb TMV promoted invasion of PC3 cells, this effect being dependent on the concentration of TMV used to stimulate fibroblasts, and returned to basal level when MV were eliminated from CM as described above. These results indicate that MV stemming from fibroblasts in response of TMV were essential to induce both migration and invasion of PC3 cells.

CX3CR1 is involved in the control of migration and invasion of PC3 cells toward CM of stimulated fibroblasts. Regarding the implication of CX3CR1-CX3CL1 axis, we have verified the expression of CX3CR1 in PC3 and LnCaP cells, fibroblasts. THP-1 cells that constitutively express CX3CR1 were included in the experiments as positive control. The presence of CX3CL1 was assessed in MV isolated from CM of fibroblasts treated with TMV_{PC3 Act.D} (MV CM-Fb TMV) toward control (22). As shown in Fig. 6A,-a, CX3CR1 was detected in PC3 cells and fibroblasts but not in LnCaP cells. On the other hand, CX3CL1 was detected in MV from CM of fibroblasts treated by TMV and in TMV_{PC3 Act.D} (Fig. 6A,-b). By neutralizing CX3CL1 in CM-Fb TMV or CX3CR1 in cancer cells with a specific Ab, migration of PC3 cells toward CM-Fb TMV was partially but significantly inhibited (Fig. 6B). No basal migration of PC3 cells in the presence of anti-CX3CL1 or anti-CX3CR1 was observed. In addition, no migration of LnCaP cells was induced by CM-Fb (data not shown), probably due to the absence of CX3CR1 expression in these cells. This suggests that CX3CR1 is implicated in the control of migration of PC3 tumor cells, when activated fibroblasts treated with TMV respond through the shedding of MV presenting a chemotactic potential for PC3 cells.

Here, we propose a part for MV derived from the plasma membrane of tumor cells in the transfer of biological information between tumor and normal cells as a mode of communication within the tumor microenvironment. Evidence is provided that such TMV have a potential to alter normal cell function, fibroblasts here. TMV promote MMP-9 expression and ERK1/2 phosphorylation in stimulated fibroblasts that become procoagulant, through increase of TF activity expression and PS exposure (Supplementary Fig. S1A and S1B), and acquire partial chemoresistance. Furthermore, TMV induce migration of fibroblasts and release of MV that in turn promote migration and invasion of cancer cells. Finally, this effect is mediated by CX3CR1, at least partially.

MMP are implicated in extracellular matrix degradation, facilitating cancer cell migration and invasion, and favoring a crosstalk between tumor and stromal cells (23). MMP vehicled by membrane MV shed from tumor and endothelial cells (24–26) may precisely represent a way to interact with the microenvironment. We have therefore assessed the expression of members of this family of protease in TMV. TMV stemming from PC3 cells are enriched in MMP-9 and -14, whereas enrichment in MMP-13 was detected in TMV from LnCaP cells (27).

Treatment of fibroblasts by TMV induced dose-dependent ERK1/2 phosphorylation, known to regulate fundamental processes including proliferation, differentiation, and cell survival (28, 29). TMV from PC3 cells promoted higher migration capacity of fibroblasts than TMV from LnCaP cells, probably in relation with the cellular origin and/or enrichment in MMP-14 observed in TMV from PC3 cells. The implication of the ERK1/2 pathway in pro–MMP-9 expression has been reported by others (30, 31). Our results are in agreement with those obtained with platelet MV, described to induce increase of ERK1/2 phosphorylation and MMP-9 release in fibroblasts. The authors concluded that platelet MV participate in the invasiveness of breast cancer cells (32). In addition, it has been suggested that platelet MV could contribute to the formation of new blood vessels in growing tumor, via the activation of mitogen-activated protein kinase pathway (25). Another molecule involved in cell migration and activation is PSMA, expressed by normal and neoplastic prostate epithelium, as negative regulator of invasion capability (18). PC3 and LnCaP cells have inverse invasion capability, possibly correlated to PSMA expression at the plasma membrane.

In the context of drug efflux, through the release of TMV for instance, the latter could play a relevant role in the attenuation of sensitivity to drug. In fact, there is a strict correlation between MV shedding, as a mechanism of drug expulsion, and anticancer drug resistance (33). We have observed that TMV_{PC3 Act.D} were able to decrease Act.D-induced apoptosis in fibroblasts, whereas TMV_{PC3 NT} had no effect (data not shown). In addition, ERK1/2 pathway is thought to be involved in cell proliferation and protect cells from apoptosis in normal and cancer cells (20), contributing to their drug-resistant nature (34). In the presence of ERK inhibitor, U0126, the effect of TMV on induced-apoptosis of fibroblasts was abolished, suggesting that ERK pathway is activated by TMV to support cell survival.

The interactions between tumor cells and the stroma, including fibroblasts, were described in the establishment of a tumor microenvironment favorable to cancer progression. The latter is thought to be enriched in MV released by cancer cells, platelets, monocytes, and lymphocytes infiltrating the tumor tissue (35). Fibroblasts are associated with tumor cells and, when activated, are believed to play a central role in migration and invasion of cancer cells (7). However, the mechanisms leading to fibroblast activation are far from being fully understood. Because fibroblasts represent the main constituents of the tumor stroma, it was relevant to assess the biological effects of TMV on this cell type. MV isolated from CM of TMV-treated fibroblasts were capable to promote the migration and invasion of cancer cells, suggesting that the activated state of fibroblasts can support the pathogenicity of cancer through the release of MV. CX3CR1 seems to play a part in this process. The association with its cognate ligand could regulate adhesion, migration, and survival of these cells (12). Because CX3CR1 expression was verified in human prostate cancer cells and CX3CL1 is present in MV isolated from CM of TMV-treated fibroblasts, the blockade of CX3CR1 or CX3CL1 by specific Ab induced partial, but significant, reduction of tumor cell migration. Even PC3 TMV contain CX3CL1 likewise fibroblast MV (Fig. 6A,-b), fibroblasts released a large proportion of MV in the CM (Fig. 1C) after TMV treatment. The difference of CX3CL1-dependent PC3 migration seems related to the number of fibroblast MV bearing CX3CL1, interacting with CX3CR1 on PC3 cells to a greater extent than with TMV. This is confirmed by the absence of PC3 basal migration in the presence of anti-CX3CL1, and by the lack of effect of TMV on their own (Fig. 5A -d). This suggests CX3CR1/CX3CL1 involvement in cancer cell migration toward CM of TMV-treated fibroblasts, together with other soluble factor(s). This is in agreement with a study where fractalkine increased integrin avidity to their ligands on monocytic CX3CR1⁺ cells (36).

Consistent with a model of chemokine receptor heterodimerization (37), transactivation of growth factor receptor by chemokines has been reported to induce prostate cancer cell invasion (38). Moreover, the membrane-bound form of fractalkine, vehicled by lipid vesicles, was observed to increase natural killer cell activation, compared with the soluble form (39). Due to MV potency as vehicles of selectively sorted biological signals, it is tempting to speculate on their differential effects elicited in target cells possibly resulting from receptor cooperation.

Altogether our results support the evidence that fibroblasts and MV are key actors within tumor environment to sustain and/or promote cancer pathogenesis.

In conclusion, we postulate that human TMV are important mediators of intercellular crosstalk between cancer cells and normal fibroblasts. The present results emphasize the role of MV in cancer progression and in mechanisms attributable to membrane-associated mediators in addition to truly soluble factors.

No potential conflicts of interest were disclosed.

Grant support: La Ligue Contre le Cancer (Comité du Haut-Rhin) and an institutional grant from Institut National de la Sante et de la Recherche Medicale.

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.

D. Castellana is recipient of a doctoral fellowship from La Ligue Contre le Cancer (section 68 Comité du Haut-Rhin).