Metastatic Properties of Prostate Cancer Cells are Controlled by VEGF

Abstract

Mechanisms of metastasis, the major complication of prostate cancer, are poorly understood. In this study, we define molecular mechanisms that may contribute to the highly invasive potential of prostate cancer cells. Vascular endothelial growth factor (VEGF), its receptors (VEGFRs), and α₅β₁ integrin were expressed by prostate cancer cells in vitro and by prostate tumors in vivo, and their expression was elevated at sites of bone metastasis compared to original prostate tumor. VEGF, through interaction with its receptors, regulated adhesive and migratory properties of the cancer cells. Specifically, the highly metastatic prostate cancer cell subline LNCaP-C4-2 showed a decreased adhesive but an enhanced migratory response to fibronectin, a ligand for α₅β₁ integrin, compared to its nonmetastatic counterpart. A similar pattern was also observed when bone sialoprotein was used as a ligand in migration assays. Increased migration of metastatic prostate cancer cells to fibronectin and bone sialoprotein was regulated by VEGF via VEGFR-2. Tumor suppressor PTEN was involved in control of VEGF/VEGFR-2 stimulated prostate cancer cell adhesion as well as proliferation.

Keywords:

• Extracellular matrix
• fibronectin
• integrin
• prostate cancer
• VEGF

INTRODUCTION

The five-year survival rate of prostate cancer confined within the capsule at the time of diagnosis is nearly 100%. Unfortunately, once metastasized, the five-year survival rate drops to 31%. Given these data, it is surprising that the number of studies focusing on prostate cancer metastasis remains relatively low.

Specific extracellular matrix proteins play a key role in cancer cell invasion, as well as in cell survival within the new matrix environment (9). By mediating tumor cell attachment, bone-matrix proteins essentially control bone invasion (36). Included in the category of bone-matrix proteins are the highly phosphorylated RGD-motif protein bone sialoprotein (BSP) and the broadly distributed matrix constituent fibronectin (33). Interestingly, many malignant tumors, together with prostate carcinomas, express BSP (1, 26, 40). Moreover, BSP serves as a marker of calcifying tumors (40), and greater BSP expression correlates with an increased risk of metastasis to bone tissue (2). The recent demonstration that osteotropic cancers express BSP further suggests that this bone matrix protein may be involved in the preferential seeding to and growth of metastatic cells within bone (26).

The integrin family of adhesion receptors mediates cell-matrix interactions (25, 35). Integrins are heterodimers, composed of one of 18 alpha subunits and one of eight beta subunits, that combine in a restricted fashion to form noncovalent complexes (23). Each member, then, has the capacity to recognize multiple ligands; it is not uncommon for a single integrin to have a repertoire of five or more ligands (31). Integrin-ligand interactions not only mediate cell adhesion, spreading, and migration, but also trigger intracellular signaling cascades, thereby regulating responses ranging from apoptosis to gene expression.

Fibronectin-α₅β₁ integrin interactions are of particular importance for prostate cancer cell migration, invasion, and metastasis (3, 34). Fibronectin functions as a chemotaxic agent, promotes cell adhesion to the extracellular matrix, and controls cell proliferation. Peptide and antibody inhibitors of fibronectin are effective inhibitors of metastasis and potentially important reagents for the study of cancer (22). BSP is recognized by several integrins, including α_vβ₃, α_vβ₅, and α₅β₁ (38). RGD-containing peptides inhibit these interactions and block cell adhesion to BSP (28).

Neoplastic transformation is also associated with growth factor level changes within the tumor microenvironment (10). One of the most important growth factors, VEGF, has been implicated in prostate carcinogenesis and metastasis and identified as a potential target for innovative anticancer therapy (18, 19). VEGF is highly expressed by prostate cancer cells (16, 17), and this expression correlates with increasing grade, vascularity and tumorogenicity (15, 41). In several animal models, neutralizing anti-VEGF antibodies show encouraging inhibitory effects on solid tumor growth and metastatic dissemination (4, 29). Hormone ablation therapy may also reduce VEGF production (37, 39).

Previously reported studies describe the development of the LNCaP model of human prostate cancer progression (42). One of the LNCaP lineage-derived human prostate cancer cell lines, LNCaP-C4-2, is androgen-independent and highly tumorogenic, with a proclivity for bone metastasis (39). This experimental human prostate cancer model provided an opportunity to study the mechanisms underlying human prostate cancer metastasis. In this study, we define molecular mechanisms that may contribute to the highly invasive potential of prostate cancer cells. We demonstrated the expression of VEGF/VEGFRs and integrins on both cell lines as well as in tissue samples of primary prostate cancer patients and bone metastases. Using the parental LNCaP cell line and its metastatic variant, LNCaP-C4-2, we performed detailed comparisons of their adhesive and migratory responses to fibronectin and BSP and demonstrated the role of VEGF in regulation of these responses. Finally, we established the role of tumor suppressor PTEN in the proliferative and adhesive properties of metastatic LNCaP-C4-2 cells.

MATERIALS AND METHODS

Cell Lines and Materials

The following cell lines were used: LNCaP, a lineage-derived human prostate cancer cell line, and LNCaP-C4-2, an androgen-independent and highly tumorogenic prostate cancer cell line with a proclivity for bone metastasis (39). The cell lines were maintained in RPMI 1640 medium supplemented with 10% FBS. Chemicon International provided anti-α₅β₁ antibodies. Likewise, R&D Systems provided anti-VEGF and anti-VEGFR-2 antibodies.

Flow Cytometry

To analyze cell surface integrin expression, the cells were incubated with primary antibodies (15 μg/ml) or with a mouse IgG as a negative control for 60 min on ice. The cells were then washed by centrifugation and incubated with FITC-goat anti-mouse IgG (for integrins) or with FITC-goat anti-rabbit IgG (for VEGF receptors) (Molecular Probes, Eugene, OR) at 25 μg/ml. Flow cytometry analysis was performed using a FACSCalibur and the data was analyzed using the CellQuest software program (version 1.2).

Immunofluorescence Microscopy

The cells were grown on fibronectin-coated cover slips (Fisher Scientific) for 48 h, washed three times with ice-cold PBS, and fixed with 4% paraformaldehyde for 30 min. The cover slips were incubated with blocking solution containing 10% goat serum and 2% BSA in PBS, and then with a primary antibody to VEGFR-2 diluted by blocking solution at 4°C overnight. After washing with PBS, the cells were incubated with a FITC-labeled secondary antibody. Cover slips were then mounted with DAPI medium. The slides were analyzed under digital fluorescence microscopy.

Immunohistochemistry

Formalin-fixed and paraffin-embedded tissue blocks of normal prostate, prostate cancer, and bone metastasis were cut into 6 μm sections. The sections were stained with monoclonal antibody to α₅β₁ integrin (Chemicon International, Temecula, CA) and polyclonal antibodies against VEGFR-2 (Santa Cruz Biotechnology, Santa Cruz, CA) and VEGF (Sigma) using an antigen retrieval technique previously described (5). The signal was visualized using 0.05% 3.3′-diaminobenzidine tetrahydrochloride (DAKO) in 0.05 M Tris buffer (pH 7.6) containing 0.003% hydrogen peroxide. The sections were counterstained with hematoxylin (Vector). The sections were then examined using an OLYMPUS microscope, and representative areas were photographed using an OLYMPUS digital camera. Quantitative data for the intensity of VEGFR-2 staining was obtained using Image-Pro Plus. Eight paired samples for primary prostate cancer and bone metastasis were analyzed. For each sample, four fields were quantified. The intensity of VEGFR-2 expression was then normalized to that of normal prostate tissue. The significance was calculated using two sample t test in Excel.

Cell Adhesion Assays

Cell adhesion assays were performed as previously described (8). LNCaP and LNCaP-C4-2 cells were harvested and labeled with Calcein AM (Molecular Probes). The cells were preincubated with or without blocking antibodies. Cell suspensions were then added immediately to the fibronectin-coated wells. At select times (40–60 min), the wells were gently washed three times with DMEM/F12 by plate inversion. Adherent cells were quantified in a Fluorescence Multi-Well Plate Reader (Perkin Elmer, Boston, MA) and examined microscopically.

Cell Migration Assays

Cell migration assays were performed using Transwell plates (8 μm pore size) as previously described (7). Fibronectin was diluted to a selected concentration, and 10 μl of the solution was placed on the lower surface of a polycarbonate filter and air-dried. Migration was then quantified by performing microscopic cell counts at 200 × on 10–12 random fields in each well.

Transfection of PTEN cDNA in LNCaP-C4-2 Cells

The cDNA encoding human PTEN was kindly provided by Dr. C. Eng (Ohio State University, Columbus, OH). PTEN was recloned into plRES2-EGFP bicistronic vector from Clontech either in a sense or in an antisense orientation. The cells were transiently transfected with PTEN and anti-PTEN plasmids (2 μg/ml each) by using lipofectamine plus reagent (Invitrogen). After 24–48 hr of transfection the PTEN and anti-PTEN-GFP expressing cells with a mean of fluorescence intensity (MFI) of > 40 (green channel), were selected by flow cytometry and further analyzed in a proliferation assay.

Proliferation Assay

Proliferation assays were performed as previously described (20). Briefly, cells were maintained in 1% serum for 20 hr prior to the experiment. Trypsinized cells were distributed at the concentration of 2 × 10⁵ into 96-well microtiter plates and labeled with 1 μCi of ³H-thymidine per well. The wells were then incubated in the presence or absence of growth factors or blocking antibodies. After 24–48 hr, the cells were washed, and radioactivity was precipitated with TCA and quantified by scintillation.

Adhesion Assay with PTEN Transfected Cells and Fluorescence Microscopy

Adhesion assay was performed as described above and quantified microscopically using a Leica fluorescence microscope. The numbers of adherent cells transfected with PTEN, as indicated by the green color of GFP, were counted and the representative fields were photographed.

RESULTS

VEGF Receptors and Integrin Expression on Prostate Cancer Cells

Studies of growth factor receptor and integrin expression on prostate cancer cells assessed VEGFR and integrin expression profiles on LNCaP, a widely utilized prostate cancer cell line, and LNCaP-C4-2, a more aggressive and metastatic subline. The mean of fluorescence intensity (MFI) on these two cell lines is shown in Figure 1A. VEGFR-1 and VEGFR-2 were similarly expressed by each cell line. Figure 1B shows the fine, punctate staining of virtually all LNCaP-C4-2 cells for the major functional VEGF receptor, VEGFR-2. Expression levels of VEGFR-2 in normal prostate tissue were compared to that of primary tumor localized within the prostate and tumor metastasized to bone (matched pairs of tissue samples were obtained from the same patients). An example of the paired specimens is shown in Figures 1C, 1E, and 1G. In normal prostate tissue, VEGFR-2 expression was observed only in endothelial cells lining blood vessels (Figure 1C). The staining of epithelial cells was negligible in normal tissue (Figure 1C); however, it was substantially increased in prostate cancer samples (Figure 1E) and further increased in bone metastases (Figure 1G). The intensity of VEGFR-2 staining was 4 ± 1.1 units on prostate tumor cells and 21.8 ± 4.2 units within bone metastasis sites. The data indicate that VEGFR-2 is expressed in prostate tumors in vivo and, more importantly, the extent of its expression is significantly increased (5.5 fold, p = 0.045) on tumor cells at sites of bone metastasis as compared to original tumor. VEGF has an expression pattern similar to that of VEGFR-2. VEGF was expressed at higher levels within the bone metastasis sites (Figure 1J) as compared to original tumor (Figure 1I) (normal prostate tissue not shown).

1 Expression of VEGF/VEGF receptors and integrins. (A) Cells were stained with specific mAbs for 60 min on ice, washed by centrifugation, incubated with secondary antibodies, washed again and surface expression was analyzed by FACS. Fluorescence in the presence of secondary antibodies was subtracted. (B) LNCaP-C4 cells were fixed and stained for VEGFR-2 expression (second antibodies were labeled with FITC). Nuclei were stained with DAPI. (C-H) Tissue sections of normal prostate (C&D), prostate tumor (E, F), and bone metastasis (G, H) were stained with antibodies against VEGFR-2 (C, E, G) and α₅β₁ (D, F, H), respectively, then with secondary abs (ABC kit from Vector). (I and J). Sections of prostate cancer (I) and bone metastasis (J) were stained with primary antibody against human VEGF, then with secondary antibody (Vector). The signal was visualized by DAB (Vector). Scale bar = 20 μm. (See Color Plate I).

I CELL COMMUNICATION & ADHESION VOLUME 11, NUMBER 1. COLOR PLATE I. See J. Chen et al., Figure 1.

In addition, integrin expression levels of α_Vβ₃, α_Vβ₅, and α₅β₁ on the prostate cancer cell lines were analyzed (Figure 1A). Both α _V integrins were present and expressed at similar levels on LNCaP and LNCaP-C4-2 cells. Integrin α₅β₁ also was present on both cell lines and its expression level was substantially higher than that of the α_V integrins. Because of this, intensity of α₅β₁ integrin expression in bone metastasis was compared with that of the primary tumor. An example is shown in Figures 1D, 1F and 1H. The α₅β₁ integrin was expressed on epithelial cells in normal prostate tissue (Figure 1D). However, its expression was substantially greater in primary prostate cancer samples (Figure 1F), and further increased within bone metastasis sites (Figure 1H).

Adhesive Responses of Prostate Cancer Cells

Since fibronectin receptor α₅β₁ was the major integrin expressed on the LNCaP and LNCaP-C4-2 cells, a functional analysis of its activity was performed. Adhesion of the two prostate cancer cell lines to fibronectin is shown in Figure 2. While the LNCaP cell line exhibited extensive adhesion to fibronectin, LNCaP-C4-2 exhibited an almost 3-fold drop in adhesion to fibronectin. In both cases, however, adhesion to fibronectin was mediated by α₅β₁, as shown by the decrease in adhesion when a monoclonal antibody specific for the integrin was added. As shown in Figure 2, blocking anti-VEGFR-2 antibody that serves to neutralize the effect of endogenous VEGF had little effect on adhesion of LNCaP cells (Figure 2A) but produced substantial inhibition of fibronectin adhesion of LNCaP-C4-2 cells (Figure 2B). The extent of inhibition with the LNCaP-C4-2 cells was about 50%. The neutralization of VEGF with antibodies produced only 20% inhibition of adhesion, due to the relatively short time of the incubation. Similar results were obtained using anti-VEGF neutralizing antibodies (data not shown). These data suggest that VEGF produced by tumor cells and engaged by VEGFR-2 may effect α₅β₁ functional activity in LNCaP-C4-2 cells. Addition of exogenous VEGF had minimal effect on adhesion of both LNCaP and LNCaP-C4-2 cells to fibronectin. Although the two key receptors, VEGFR-2 and α₅β₁, are present on the LNCaP cells, the pathway does not appear to be operative. Inspection of the quantitative aspects of these data indicates a major difference in the adhesive response of LNCaP and LNCaP-C4-2 cells. LNCaP-C4-2 cells were substantially less adhesive to fibronectin than LNCaP cells.

2 Adhesion to fibronectin. LNCaP (A) and LNCaP-C4-2 (B) cells were harvested, labeled with calcein, preincubated with anti-VEGF, anti-VEGFR-2, or anti-α ₅β₁ antibody (15 μg/ml each), then were added to fibronectin-coated wells (10 μg/ml). After 1 h incubation, wells were washed and adhesion was measured. Nonspecific adhesion to BSA-coated wells was subtracted. The data shown are means ± SD of triplicates in each experiment and are representative of three separate experiments. Statistical analysis was performed using t-test, symbol (*) indicates p < 0.05 and (**) indicates p < 0.01.

Migratory Activities of Prostate Cancer Cells

The quantitative response of the two cell lines with respect to migration was also different. As shown in Figure 3A, LNCaP-C4-2 cells exhibited greater migration towards fibronectin as compared to LNCaP cells. The migration of the LNCaP-C4-2 cells towards fibronectin was α₅β₁-dependent as indicated by the inhibition of migration in the presence of α₅β₁ blocking antibodies. Migration of LNCaP-C4-2 cells was inhibited in the presence of anti-VEGF neutralizing antibodies as well as VEGFR-2 specific blocking reagents. Both reagents provided more than 60% inhibition of cell migration. Thus, the VEGF-dependent stimulation process, mediated by VEGFR-2, controls the migratory behavior of the LNCaP-C4-2 cells.

3 Migration to fibronectin (A) and to bone sialoprotein (B). The lower surfaces of filters in Transwell chambers were coated with fibronectin or bone sialoprotein (BSP) (100 ng/well), and 5 × 10⁵ cells were placed in the upper chamber, with or without anti-α₅β ₁, anti-VEGF or anti-VEGFR-2 blocking antibody (15 μg/ml each). Migration was quantified by performing microscopic cell counts of 10–12 random fields at 200× power. The data shown are means ± SD of triplicates in each experiment and are representative of three separate experiments.

We also assessed the migration of LNCaP-C4-2 cells to bone sialoprotein (BSP), a specific component of bone matrix, in the presence of anti-VEGF neutralizing antibodies as well as anti-VEGFR-2 reagents. Figure 3B shows that anti-VEGF antibodies and VEGFR-2/Fc chimera significantly inhibited migration of prostate tumor cells to BSP. The data indicate that VEGF contributes to the highly migratory phenotype of metastatic prostate tumor cells not only to fibronectin but also to the bone-specific matrix proteins exemplified by BSP. The data summarized above suggest that VEGFR-2 is engaged and maintains the LNCaP-C4-2 cells in a highly migratory state.

Role of PTEN and VEGF in Prostate Cancer Cell Proliferation

It is well established that the recognition of extracellular matrix by cancer cells influences their proliferation and survival (14). Since inhibition of VEGF significantly affected adhesive and migratory responses of metastatic prostate cancer cells, we next assessed the role of endogenous VEGF in the proliferative activity of these cells. The effects of the anti-VEGFR-2 on cell proliferation supported the concept of an active VEGF/VEGFR-2 autocrine loop (Figure 4). The blocking antibody to VEGFR-2 inhibited thymidine uptake by the LNCaP-C4-2 cells by almost 50% (Figure 4). Interestingly, a similar extent of inhibition was achieved by reexpression of the tumor suppressor protein PTEN in prostate cancer cells (the LNCaP-C4-2 cells similar to the other prostate cancer cells possess a frame-shift mutation in PTEN gene). The approach in which the cells were transfected with a bicistronic vector containing PTEN cDNA, either in a sense or an antisense orientation, and GFP was employed. In the antisense orientation, the transfected cells maintained their high spontaneous proliferation; however, the proliferative response was markedly blunted when cells were transfected with PTEN in a sense orientation. In this suppressed state, VEGF or anti-VEGFR-2 had negligible effects on proliferation. Thus, PTEN appears to play a central role in the proliferative response of LNCaP-C4-2 cells to VEGF through an engagement of VEGFR-2.

4 Prostate cancer cell proliferation. LNCaP-C4-2 was transfected with PTEN in sense or antisense orientation in the plRES2-EGFP vector. LNCaP-C4-2 proliferation in response to VEGF-A165 (20 ng/ml), 15% FCS, and anti-VEGFR-2 antibody was measured by ³H-thymidine uptake (24 h time point).

Role of PTEN in Cell Adhesion to Fibronectin

To investigate the role of PTEN in the regulation of prostate cancer cell adhesion, LNCap-C4-2 cells were transiently transfected with bicistronic vector containing GFP and PTEN in either a sense or an antisense orientation. 24 hr after transfection, 11.2 ± 5.3% of cells were expressing anti-PTEN (based on green fluorescence) and only 6.4 ± 3.6% of cells were positive for PTEN, indicating the negative effect of PTEN on cell proliferation and survival (Table 1). Control LNCaP-C4-2 cells expressing anti-PTEN showed extensive adhesion to fibronectin with subsequent spreading (Figure 5C (phase contrast) and Figure 5E (GFP fluorescence)). Among cells adherent to fibronectin, anti-PTEN transfected cells were 13.1 ± 6%, which is similar to that in suspension. In contrast, only a few cells expressing PTEN in a sense orientation were identified upon adhesion to fibronectin (less than 1% of the total number of cells counted) as shown in Table 1 and in Figures 5A and 5B (phase contrast) and Figures 5D and 5E (GFP fluorescence)). These results demonstrate that PTEN is able to down-regulate integrin-dependent adhesive responses in prostate cancer cells.

Effect of PTEN on cell adhesion

	%of GFP-positive cells in suspension	%of GFP-positive cells upon adhesion to Fn
antisense PTEN (control)	11.2 ± 5.3% (n = 3)	13 ± 6%
PTEN	6.4 ± 3.6% (n = 3)	0.8 ± 0.5%

The LNCaP-C4-2 cells were transiently transfected with bicistronic vector containing GFP and PTEN either in a sense or in an antisense orientation. 24 hr after transfection, percentages of GFP-positive cells with a mean fluorescence intensity (MFI) of >30 (green channel) were assessed by flow cytometry. The data shown are means ± SD of three experiments. Transfected cells were harvested, and the same number of cells were plated in the chambers coated with fibronectin and incubated for 1 hr. The cells were washed twice with PBS, covered with a coverslip, and analyzed using a Leica fluorescence microscope. The numbers of cells transfected with PTEN or anti-PTEN, as indicated by green color of GFP, were counted in 12 randomly selected fields and expressed as a percent of the total number of cells counted.

5 PTEN-Mediated adhesion. LNCaP-C4-2 cells transfected with PTEN in either a sense or an antisense orientation were plated in fibronectin-coated chamber and incubated at 37°C at 5% CO₂-air for 1 h. The adhesive activity of LNCaP-C4-2 transfected with PTEN cDNA as indicated by GFP were analyzed by fluorescence microscopy. The Images by phase contrast are shown in the left column (A, B, and C). The fluorescence micrograph of the same field was shown in the right column (D, E and F). The top two pairs are for sense PTEN, and the bottom pair is for antisense cDNA. Scale bar = 10 μm. (See Color Plate II).

II CELL COMMUNICATION & ADHESION VOLUME 11, NUMBER 1. COLOR PLATE II. See J. Chen et al., Figure 5.

DISCUSSION

The study demonstrated the following: (1) VEGF, VEGFRs and α₅β₁ integrin were expressed by prostate tumor cells in vitro and in vivo, creating the possibility of a VEGF autocrine loop, (2) expression levels of VEGF, VEGFRs, and α₅β₁ were elevated at the sites of bone metastasis compared to the original prostate tumor, (3) metastatic prostate cancer cell lines exhibited decreased adhesive but enhanced migratory properties to extracellular matrix proteins compared to nonmetastatic controls, (4) VEGF, through VEGFR-2, modulated the migratory responses to fibronectin and bone sialoprotein, and (5) VEGFR-2 and PTEN were involved in the regulation of prostate cancer cell proliferation as well as cell adhesion to fibronectin.

Prostate tumors, like most tumors, over express VEGF, thereby promoting the development of tumor neovascularization (17, 24). As we reported previously, the metastatic prostate cancer cell line LNCaP-C4-2 secretes significantly higher amounts of VEGF compared to its nonmetastatic counterpart LNCaP (12). In this study, we observed an increased expression of VEGF at the sites of bone metastasis as compared to the original tumor localized within the prostate, which is in agreement with published data (17). Based on FACS analysis and immunofluorescence results in vitro, as well as immunohistochemistry staining of tissue samples obtained from patients, prostate tumor cells also express high levels of the major VEGF receptor, VEGFR-2. For the first time, we demonstrate that both VEGF and VEGFR-2 expression levels were significantly higher on tumor cells at the sites of bone metastasis as compared to the original prostate tumor from the same individual. This finding implies that VEGF/VEGFR-2 is involved in the process of metastasis development.

We have shown previously that in addition to its paracrine role, VEGF serves an important autocrine function for cancer cells by regulating the activity of α_Vβ₃ integrin (6, 12). The present study demonstrated that VEGF-dependent autocrine loop controls the recognition of the major matrix component, fibronectin, by α₅β₁ integrin on metastatic cells LNCaP-C4-2. Using fibronectin as a substrate, adhesion and migration of LNCaP-C4-2 was inhibited as a result of VEGF or VEGFR-2 neutralization. Interestingly, an addition of VEGF did not further modulate adhesive activities of prostate cancer cells. It is known that the effects of VEGF strongly depend upon the concentration of this growth factor, therefore, the cellular responses are often blunted at the high concentrations (6). Nonmetastatic cells demonstrated tight adhesion and a minimal rate of integrin-dependent migration to fibronectin. The inverse correlation between adhesion and cell motility has been characterized previously (30). We can now apply this phenomenon to explain the greater migratory behavior of metastatic tumor cells.

One of our most intriguing findings is that the migratory response of metastatic, but not nonmetastatic prostate cancer cells, was affected by VEGF autocrine stimulation. This stimulation might be responsible for the high motility of tumor cells within the matrix that contributes to the development of metastasis. Moreover, VEGF supported the high proliferative rate of prostate cancer cells, which could be inhibited by VEGF or VEGFR-2 neutralization. Alternatively, proliferation of prostate cancer cells was inhibited by reexpression of intracellular mediator, tumor suppressor PTEN, which is often mutated in prostate cancer (21). Our results are in agreement with the study reported by Davies et al. (11) demonstrating that PTEN expression in alternative prostate cancer cell line, PC3 cells, inhibited cell proliferation in vitro and significantly reduced tumor size as well as the development of metastases in vivo. Interestingly, an effect of PTEN expression in PC3 prostate cancer cells on metastasis was greater than on tumor growth in general (11). Based on our previously published observations, PTEN is able to negatively regulate α_Vβ₃ integrin-dependent functions, including cell migration (6). Now we have demonstrated that PTEN reexpression inhibits the adhesive activity of prostate cancer cells to fibronectin mediated by integrin α₅β₁, implying the recognition of extracellular matrix by prostate cancer cells is controlled by PTEN. Since the process of matrix recognition is the key component of the metastasis development, PTEN might be a very attractive target for antimetastatic strategies.

Thus, VEGF-dependent autocrine stimulation targets tumor cells and may contribute to prostate cancer growth autonomy and metastatic properties. We have already demonstrated that the VEGF autocrine loop controls the α_Vβ₃ integrin activation state on melanoma cells and prostate cancer cells (6, 12). Although the existence of a growth-factor dependent autocrine loop becomes a common phenomenon in cancer biology (13, 27), it seems to be involved not only in cell proliferation and survival but also in the recognition of extracellular matrix components, which, in turn, might influence the metastatic activities of prostate cancer cells.

ACKNOWLEDGMENT

This work was supported by CAPCURE Foundation and NIH grant DK060933 to T. Byzova.