α5β1 Integrin interacts with the PHSRN sequence of plasma fibronectin, causing constitutive invasion by human prostate cancer cells. Inhibition of this process reduces tumorigenesis and prevents metastasis and recurrence. In this study, naturally serum-free basement membranes were used as in vitro invasion substrates. Immunoassays were employed to dissect the roles of focal adhesion kinase (FAK), phosphatidylinositol 3′-kinase (PI3K), and protein kinase Cδ (PKCδ) in α5β1-mediated, matrix metalloproteinase-1 (MMP-1)–dependent invasion by metastatic human DU 145 prostate cancer cells. We found that a peptide composed of the PHSRN sequence induced rapid FAK phosphorylation at Tyr397 (Y397), a site whose phosphorylation is associated with kinase activation. The technique of RNA silencing [small interfering RNA (siRNA)] confirmed the role of FAK in PHSRN-induced invasion. PHSRN also induced the association of the p85-regulatory subunit of PI3K with FAK at a time corresponding to FAK phosphorylation and activation, and maximal PI3K activity occurred at this same time. The necessity of PI3K activity in both PHSRN-induced invasion and MMP-1 expression was confirmed by using specific PI3K inhibitors. By employing a specific inhibitor, Rottlerin, and by using siRNA, we also found that PKCδ, a PI3K substrate found in focal adhesions, functions in PHSRN-induced invasion. In addition, the induction of MMP-1 in PHSRN-treated DU 145 cells was shown by immunoblotting, and the role of MMP-1 in PHSRN-induced invasion was confirmed by the use of blocking anti-MMP-1 monoclonal antibody. Finally, a close temporal correspondence was observed between PHSRN-induced invasion and PHSRN-induced MMP-1 activity in DU 145 cells. (Cancer Res 2006; 66(16): 8091-9)
Integrins are heterodimeric receptors that transmit signals between extracellular matrix ligands and intracellular signal transduction pathways (1) to support adhesion, invasion, survival, and differentiation. The α5β1 integrin interacts with the PHSRN sequence of fibronectin cell-binding domain fragments to stimulate invasion during wound healing (2). Because epithelial cells, fibroblasts, and endothelial cells express the invasion-inhibitory α4β1 integrin as well as α5β1, fibronectin fragmentation is required for invasion induction (3). Integrins are also very important in cancer progression (4). For example, the α5β1 fibronectin receptor is specifically up-regulated in late-stage tumors as a consequence of epithelial to mesenchymal transition induction and establishment of the invasive phenotype (5), whereas α4β1 may be down-regulated, resulting in constitutive, fibronectin-dependent invasion (6).
To evaluate the contribution of plasma fibronectin (pFn) in metastatic invasion, we used naturally serum-free, selectively permeable (7) basement membranes of sea urchin embryos (SU-ECM) as in vitro invasion substrates. SU-ECM have been shown to be free of background invasion by unstimulated normal cells (2, 6), which can be observed when reconstituted basement membranes are used (8). In addition, SU-ECM are structurally and functionally similar to mammalian basement membranes (9). Previously, we showed that the α5β1/pFn interaction causes constitutive invasion by metastatic human prostate cancer cells and identified PHSRN as the invasion-inducing sequence (6). We have also found that matrix metalloproteinase 1 (MMP-1)–dependent invasion is caused by the interaction of the α5β1 receptors of metastatic breast cancer cells with pFn (9, 10) because, like prostate cancer (11), these cells frequently lose α4β1 from their cell surfaces, thereby becoming invasive in the presence of the pFn of all body fluids (10, 12). Hence, we explored the effects of the PHSRN sequence on intracellular signaling downstream of the α5β1/pFn interaction in prostate cancer cells.
Focal adhesion kinase (FAK) is a critical mediator of integrin-mediated migration and signaling (13). It exhibits increased kinase activity (14) and Tyr397 (Y397) phosphorylation after integrin activation (14, 15). FAK is also up-regulated in prostate cancer cell lines and in lysates of prostate tissues from patients with metastatic disease but is not increased in normal prostate or in tissues from patients with localized prostate cancer (16). Furthermore, RNA interference (RNAi) has shown that FAK plays an important role in cancer cell invasion (17).
When FAK is activated and autophosphorylated at Y397, it has been shown to bind the SH2 domain of phosphatidylinositol 3′-kinase (PI3K; ref. 18), thereby bringing the catalytic subunit of PI3K to the membrane, where it catalyzes phosphorylation of inositol lipids at the D-3 position to form 3′-phosphorylated phosphoinositides, including phosphatidylinositol-3,4,5-trisphosphate (PIP3). Integrin activation-induced FAK/PI3K association has been shown in both platelets and fibroblasts (19). Furthermore, PI3K plays an important role in invasion by many types of cancer (20–22). Moreover, we have shown that erbB-2 oncogene overexpression down-regulates α4β1, thereby inducing PI3K-dependent invasion in transformed mammary epithelial cells and in human breast cancer, thus linking the α5β1+α4β1− integrin fibronectin receptor phenotype to constitutive, PI3K-dependent invasion (10, 23). However, the role of PI3K in α5β1-mediated prostate cancer cell invasion has not yet been investigated.
Thus, we examined the effects of the PHSRN peptide on FAK Y397 phosphorylation, on the association of FAK with the PI3K p85 subunit, and on PI3K activity in metastatic DU 145 prostate cancer cells. By using RNAi (24) to reduce FAK expression and specific inhibitors of PI3K, we also determined the effects of FAK and PI3K on PHSRN-induced invasion and MMP-1 secretion. Because protein kinase Cδ (PKCδ) is activated by the PI3K product, PIP3 (25), we also assessed its role in PHSRN-induced invasion and MMP-1 activity. Our results implicate a pathway involving FAK, PI3K, and PKCδ in α5β1-mediated invasion by metastatic prostate cancer cells.
Materials and Methods
Cell culture. DU 145 cells, originally cultured from a human prostate adenocarcinoma brain metastasis (26), were obtained from the American Type Culture Collection (Manassas, VA) and cultured as recommended. For all assays in serum-free medium, DU 145 cells were first serum-starved overnight.
Peptide synthesis. NH2-terminal acetylated, COOH-terminal amidated PHSRN, HSPNR, and LDV peptides (Ac-PHSRN-NH2, Ac-HSPNR-NH2, and Ac-LHGPEILDVPST-NH2) were synthesized, and their purities were assessed as described previously (2, 6, 9). Peptide purities were found to be 97% for Ac-PHSRN-NH2, 93% for Ac-HSPNR-NH2, and 91% for Ac-LHGPEILDVPST-NH2 (data not shown). Peptide structures were confirmed, and peptides were desalted and stored as described previously (2, 6, 9).
Invasion assays. Serum-free SU-ECM in vitro invasion substrates were prepared, and invasion assays were done as described previously (6, 9). For assays evaluating the effects of anti-MMP-1-blocking monoclonal antibody (mAb) on PHSRN-induced invasion, the concentration of Ac-PHSRN-NH2 in the serum-free assay medium was 1 μg/20,000 cells. Function-blocking anti-MMP-1 (COMY-4A2), anti-MMP-2 (CA-4001), and anti-MMP-9 (GE-213) mAb (Chemicon International, Temecula, CA; refs. 27–29) were prebound to cells at concentrations ranging from 10 to 300 μg/ml, and invasion was assayed as described above. Isotype control antibodies, IgM (BD Biosciences PharMingen, San Diego, CA) and IgG1 and IgG3 (Sigma, St. Louis, MO), were prebound to cells for invasion assay controls as described above. For determining the effect of the PI3K inhibitor, LY294002, on PHSRN-induced invasion, serum-free DU 145 cells were treated with 25 μmol/L LY294002 (Sigma) for a day before suspension, addition of PHSRN peptide, and invasion assay on SU-ECM as described above. For determining the effect of the PI3K inhibitor, wortmannin (Calbiochem, San Diego, CA), on PHSRN-induced invasion, serum-free DU 145 cells were suspended, rinsed, and treated with 10 nmol/L wortmannin for 10 minutes at 37°C before the addition of the PHSRN peptide and placement on SU-ECM in serum-free medium. For assaying the effect of the PKCδ inhibitor, Rottlerin, on PHSRN-induced invasion, serum-free DU 145 cells were treated with 5 μmol/L Rottlerin (Calbiochem) for 15 minutes at 37°C before PHSRN peptide addition and inclusion in SU-ECM invasion assays. For determining the effect of the PKCα inhibitor, Safingol, on PHSRN-induced invasion, serum-free DU 145 cells were treated with 20 μmol/L Safingol (Calbiochem) for 15 minutes at 37°C before PHSRN treatment and invasion assay as described above.
Assay for PI3K activity. PI3K activity was determined using the PI3K ELISA kit (Echelon Biosciences, Inc., Salt Lake City, UT) according to the manufacturer's instructions and as described previously (30). This kit measures PI3K activity by quantifying the conversion of phosphatidylinositol-3,4-bisphosphate (PIP2) into PIP3. Serum-free DU 145 cells were treated with Ac-PHSRN-NH2 for various times as indicated. Cell lysates were incubated for 1 hour at 4°C with anti-PI3K p85 antibody (Upstate Biotechnology, Lake Placid, NY) followed by addition of protein G-agarose beads (Roche Applied Science, Indianapolis, IN) according to the manufacturer's instructions. In these assays, the colorimetric signal is inversely proportional to the amount of PIP3 produced by PI3K activity.
MMP-1 activity assay. MMP-1 activity in adherent DU 145 cells was measured as described previously (9). When appropriate, cells were treated with LY294002, Rottlerin, or Safingol inhibitors before treatment with the PHSRN peptide as described above. Results were analyzed using Student's t test.
Immunoprecipitation and immunoblotting. DU 145 cells were cultured, serum starved, peptide treated, and lysed as described previously (9). Lysates containing 500 μg protein were immunoprecipitated using anti-FAK antiserum (Upstate Biotechnology) according to the manufacturer's instructions. Protein G-agarose beads were mixed with lysates, washed, resuspended in SDS sample buffer, and boiled to release the bound immunocomplexes. The levels of phosphorylated FAK were measured by immunoblot analysis, as described previously (9), using anti-FAK (pY397) phosphospecific antibody (Biosource International, Inc., Camarillo, CA). The blots were stripped and reprobed with anti-PI3K p85 mAb to determine p85/FAK association. To assess total FAK levels, membranes were reprobed with anti-FAK mAb. Levels of pY861 FAK were evaluated by immunoblotting using an anti-FAK (pY861) antibody (Biosource International). The effects of PHSRN peptide treatment on PKCδ Thr505 (T505) phosphorylation levels in serum-free DU 145 cells were evaluated by immunoblotting using an activation-specific, anti-PKCδ (pT505) antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and an antibody that detects total PKCδ protein (Cell Signaling Technology, Beverly, MA). The effects of PHSRN treatment on PKCα Ser657 (S657) phosphorylation levels were evaluated by immunoblotting using an activation-specific, anti-pS657 antibody (Upstate Biotechnology). The effects of peptide treatment on MMP-1 expression were analyzed by immunoblotting as described previously (9). The treatment groups were as follows: serum-free medium only, serum-free medium plus 1 μg/20,000 cells Ac-PHSRN-NH2 peptide, serum-free medium plus 1 μg/20,000 cells Ac-HSPNR-NH2 peptide, serum-free medium plus 2.5 μg/20,000 cells Ac-LHGPEILDVPST-NH2 peptide, and serum-free medium plus 1 μg/20,000 cells Ac-PHSRN-NH2 plus 2.5 μg/20,000 cells Ac-LHGPEILDVPST-NH2. Band densities were quantified using ImageJ software.1
Small interfering RNA. Two small interfering RNA (siRNA) oligonucleotides and a RNAi Starter kit were obtained from Qiagen, Inc. (Valencia, CA). The siRNA sequence used for targeting PKCδ was AAGGCTGAGTTCTGGCTGGAC (31), Genbank accession no. NM-006254 (Qiagen). The siRNA sequence for targeting FAK was from target regions 248 to 298, validated siRNA (32), Genbank accession no. NM-153831 (Qiagen). DU 145 cells were seeded, and transfections of siRNA were done according to the manufacturer's instructions. siRNA (1 μg) was transfected using 6 μL RNAiFect transfection reagent. Nonsilencing fluorescein-labeled control siRNA was used for monitoring transfection efficiency. At 24 to 72 hours after transfection, cells were switched to serum-free culture medium for 24 hours and then treated with Ac-PHSRN-NH2 or Ac-HSPNR-NH2 peptides as described above. FAK and PKCδ protein knockdown levels were assessed by immunoprecipitation and immunoblotting using anti-FAK (Upstate Biotechnology) or anti-PKCδ antibody (Sigma). The effects of FAK or PKCδ siRNA on PHSRN-induced MMP-1 expression were assessed using anti-MMP-1 polyclonal antibody (Chemicon International) with purified human MMP-1 (Chemicon International) as a positive control. Functional effects of FAK or PKCδ siRNA on PHSRN-induced DU 145 invasion were evaluated in serum-free SU-ECM invasion assays as described above.
3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), a commonly used method for measuring cellular viability by assaying mitochondrial dehydrogenase activity (33), was used to assess the viability of DU 145 cells expressing FAK or control siRNA. MTT assays were done using an in vitro toxicology assay kit (Sigma) according to the manufacturer's instructions. DU 145 cells were inoculated into 96-well plates at a density of 4,000 per well and cultured for 24 hours. Then, the cells were serum starved overnight and switched to medium without phenol red. DU 145 cells were exposed to 5 nmol/L FAK siRNA (Qiagen), 5 nmol/L nonspecific siRNA (Qiagen), or 20 μmol/L cisplatin (Sigma) for periods ranging from 24 to 72 hours. The absorbance of wells was measured using a microplate spectrophotometer (model SPECTRAmax PLUS384, Molecular Devices Corp., Sunnyvale, CA). The background absorbance at 690 nm was subtracted from the 570 nm signal.
Induction of FAK phosphorylation and PI3K p85 association by PHSRN. We have shown that the PHSRN peptide induces invasion by serum-free DU 145 human prostate carcinoma cells, as well as by various human breast cancer cell lines (6, 9). Furthermore, PI3K activity is required for α5β1-mediated invasion by transformed mammary epithelial cells (23). Thus, we compared the effects of the PHSRN and scrambled HSPNR peptides on FAK Y397 phosphorylation and on FAK/PI3K association. As shown in a typical immunoblot of lysates from PHSRN- and HSPNR-treated DU 145 cells probed with anti-pY397 FAK antibody (Fig. 1A), exposure to the PHSRN peptide induced a rapid, transient FAK phosphorylation at Y397. Significant phosphorylation occurred within 15 minutes, peaked at 30 to 60 minutes, and then began to decline. No changes in the total amounts of FAK protein were observed. In addition, FAK Y397 phosphorylation was not observed after treatment with the HSPNR control peptide, showing that PHSRN-induced FAK Y397 phosphorylation depends specifically on the PHSRN sequence. Moreover, the PHSRN peptide does not seem to induce FAK phosphorylation on Y861, a Src-dependent site, whose phosphorylation is independent of adhesion (34). Figure 1A also shows a histogram depicting the mean relative levels of pY397 FAK and total FAK, with their first SDs, which were obtained by densitometric analysis of the appropriate bands on multiple anti-FAK (pY397) immunoblots like the one shown above. Exposure to the PHSRN peptide induced rapid FAK phosphorylation at Y397, with levels peaking after 30 to 60 minutes of exposure and declining thereafter. No changes in the total amounts of FAK protein were observed. In addition, no detectable FAK phosphorylation at Y397 occurred after exposure to the scrambled peptide control. As expected from the fibronectin dependence of serum-induced invasion and the role of the fibronectin PHSRN sequence in invasion induction (2, 6, 9), similar time courses of FAK Y397 phosphorylation were induced in serum-starved DU 145 cells after exposure to serum-containing medium (data not shown).
To assess the effects of PHSRN peptide on FAK/p85 association, cultures of adherent, serum-free DU 145 cells were treated with the PHSRN or HSPNR peptides for various times. As shown in the sample blot of Fig. 1B, exposure to the PHSRN peptide induced a rapid, transient association of the PI3K p85 subunit with FAK, which peaked after 30 to 60 minutes, closely corresponding to the time of maximal FAK pY397 levels. No significant PI3K p85/FAK association was observed in response to HSPNR peptide treatment. The same time course of PHSRN-induced FAK/p85 association was also observed when anti-PI3K p85 immunoprecipitates were immunoblotted and probed with anti-FAK antibody (data not shown). Figure 1B also shows a histogram comparing the relative amounts of the FAK-associated p85 subunit of PI3K to total FAK protein in multiple anti-p85 immunoblots of anti-FAK immunoprecipitates. Multiple anti-FAK immunoprecipitates, each obtained at various times after exposure of serum-starved DU 145 cells to the PHSRN or HSPNR peptides, were analyzed on each immunoblot. Mean relative levels of FAK-associated PI3K p85 subunit, and total FAK protein, each obtained by densitometric analysis of the appropriate bands on several immunoblots, are shown with their first SDs. As in the example shown in Fig. 1A, Fig. 1B shows that exposure to the PHSRN peptide induced rapid association of the PI3K p85 subunit with FAK, which peaked after 30 to 60 minutes of exposure and declined thereafter. As indicated, no changes in the total amounts of FAK or p85 proteins were observed. Also like Fig. 1A, no detectable p85 association with FAK occurred after exposure to the scrambled peptide control. Similar time courses of FAK/PI3K p85 subunit association were induced in serum-starved DU 145 cells after exposure to serum-containing medium (data not shown).
The effect of the PHSRN peptide on PI3K activity was assayed by a competitive, colorimetric ELISA assay (30) in serum-free DU 145 cells at various times after exposure. The mean absorbances plotted in Fig. 1C are inversely proportional to the amount of PIP3 produced by DU 145 cells in response to PHSRN. Thus, PI3K activity was detectable after 15 minutes of PHSRN peptide treatment in adherent DU 145 cells. Moreover, activity peaked at ∼30 minutes, corresponding to the time of maximal PHSRN-induced pFAK/PI3K p85 subunit association, as seen in Fig. 1B. In addition, PHSRN-induced PI3K activity was transient; it decreased after 60 minutes, later becoming undetectable. To verify the specificity of the ELISA assay used to detect PI3K activity, PHSRN-induced PI3K activity was assayed in the presence of the LY294002 or wortmannin PI3K inhibitors. As expected, exposure of DU 145 cells to either LY294002 or wortmannin reduced PHSRN-induced PI3K activity to background levels as shown in Fig. 1D.
Prevention of PHSRN-induced invasion by overexpression of FAK siRNA. The functional role of FAK in PHSRN-induced invasion was assessed by overexpressing in DU 145 cells, an siRNA specific for FAK. The technique of siRNA overexpression is a well-known method for achieving specific, transient gene silencing in mammalian cells (24). An immunoblot, comparing FAK expression in DU 145 cells overexpressing either FAK siRNA or a nonspecific negative control siRNA for 2 days, is shown in Fig. 2A. Both siRNAs were present at a concentration of 1 nmol/L as indicated. In contrast to nonspecific siRNA, overexpression of FAK siRNA nearly eliminated detectable FAK expression in DU 145 cells. The effect of 24 to 72 hours of FAK siRNA overexpression on PHSRN-induced DU 145 invasion was then assessed using serum-free SU-ECM basement membranes (2, 6). Figure 2B shows the mean invasion percentages for PHSRN-treated, serum-free DU 145 cells. Two days after transfection, the percentage of invaded DU 145 cells, expressing nonspecific siRNA, was very similar to that obtained for untreated DU 145 cells, assayed in parallel, and to the value reported previously (6); however, invasion was completely prevented in DU 145 cells overexpressing FAK siRNA, suggesting that FAK activity is required for PHSRN-induced invasion. Very similar effects of FAK siRNA and nonspecific siRNA on invasion were also observed 1 and 3 days after transfection (data not shown). Because exposure to FAK siRNA has been shown to cause decreases in both viability and fibronectin-induced migration in several types of cancer cells, in addition to reducing FAK protein levels by ∼70% (35), we evaluated the effect of FAK siRNA on DU 145 viability using the MTT assay for mitochondrial dehydrogenase activity (33). As shown in Fig. 2C, exposure to FAK siRNA for 2 days had no discernable effect on DU 145 viability, whereas exposure to cisplatin, a broad-activity antineoplastic agent, significantly decreased it. Very similar results were obtained 1 and 3 days after siRNA transfection. In addition, the results of trypan blue exclusion gave results fully consistent with the MTT assays (data not shown). Thus, FAK down-regulation seems to prevent PHSRN-induced invasion without significantly decreasing DU 145 viability.
Role of PI3K in PHSRN-induced invasion. Because PHSRN induces invasion by DU 145 cells (6) and because PI3K also functions in serum-induced invasion by transformed mammary epithelial cells (23), the role of PI3K in PHSRN-induced invasion was assessed. Serum-free DU 145 cells were treated with either LY294002 or wortmannin PI3K inhibitors; then invasion was induced with the PHSRN peptide and assayed on serum-free SU-ECM basement membrane-containing invasion substrates. As shown in Fig. 2D, both PI3K inhibitors completely prevented invasion, suggesting that PI3K activity is required for PHSRN-induced invasion in DU 145 cells.
Role of PKCδ in PHSRN-induced invasion. Because the lipid products of PI3K have been shown to activate PKCδ (25), its functional role in PHSRN-induced invasion was assessed by overexpression in DU 145 cells of an siRNA specific for PKCδ. An immunoblot, comparing PKCδ expression in DU 145 cells overexpressing either PKCδ siRNA or nonspecific siRNA, 65 hours after transfection, is shown in Fig. 3A. PKCδ siRNA, but not nonspecific siRNA, substantially reduced PKCδ expression in DU 145 cells. The effect of PKCδ siRNA overexpression on PHSRN-induced invasion was then assessed in these cells using serum-free SU-ECM. Figure 3B shows the mean invasion percentages for serum-free DU 145 cells treated with the PHSRN peptide to induce invasion 65 hours after siRNA transfection. The mean percentage of invaded DU 145 cells, expressing nonspecific siRNA, was very similar to that of untreated controls (data not shown) as well as to that reported for DU 145 cells (6); however, invasion was completely prevented in DU 145 cells overexpressing PKCδ siRNA, suggesting that PKCδ activity is required for PHSRN-induced invasion.
To verify the requirement for PKCδ activity in invasion, serum-free DU 145 cells were treated with Rottlerin, a specific inhibitor of PKCδ, or with Safingol, a selective inhibitor of PKCα (36, 37). As shown in Fig. 3C, 5 μmol/L Rottlerin prevented PHSRN-induced invasion by DU 145 cells, whereas 20 μmol/L Safingol had no effect. These results suggest that PKCδ functions in PHSRN-induced invasion, whereas PKCα may not play an appreciable role. They also imply that exposure to the PHSRN peptide should result in PKCδ activation, as indicated by T505 phosphorylation in its activation loop (38), but not in the activation of PKCα, as indicated by S657 phosphorylation (39). To ascertain whether the PHSRN peptide specifically induces PI3K-dependent T505 phosphorylation of PKCδ, serum-free DU 145 cells were treated with the PHSRN peptide in the absence or presence of the PI3K inhibitor wortmannin, and levels of PKCδ pT505 and PKCα pS657 phosphorylation were evaluated by immunoblotting. As shown in Fig. 3D, PHSRN treatment of DU 145 cells significantly increases the amount of activated PKCδ, as shown by phosphorylation on T505, without affecting the amount of activated PKCα phosphorylated on S657. Furthermore, PHSRN-induced PCKδ phosphorylation on T505 seems to require PI3K, as shown by its sensitivity to wortmannin.
Role of MMP-1 in PHSRN-induced invasion by DU 145 cells. MMP-1, but neither MMP-2 nor MMP-9, is necessary for PHSRN-induced invasion by breast cancer cells and mammary epithelial cells (9). Thus, the role of MMP-1 was assessed in suspended DU 145 cells, treated with PHSRN peptide, and placed on SU-ECM invasion substrates. As shown in Fig. 4A, pretreatment with increasing concentrations of function-blocking anti-MMP-1 progressively blocked PHSRN-induced invasion by DU 145 cells, whereas the isotype control as well as elevated concentrations of blocking anti-MMP-2 or anti-MMP-9 had no effect. Thus, MMP-1 activity seems to be required specifically for α5β1-mediated, PHSRN-induced invasion by DU 145 cells.
The effect of the PHSRN peptide on MMP-1 activity, secreted into the medium by adherent DU 145 cells, was assessed by using a quantitative, colorimetric assay (40). As shown in Fig. 4B, secreted MMP-1 activity rapidly increased in the medium of DU 145 cultures in response to treatment with PHSRN peptide. Statistical analysis of the differences in MMP-1 activity between PHSRN-treated cells and untreated cells incubated in parallel cultures for the same periods of time showed P0 to be <0.0001 (data not shown). Moreover, the kinetics of MMP-1 secretion corresponded well with the time course of PHSRN-induced invasion also shown in Fig. 4B. The percentage of invaded cells was half-maximal in 1 hour and reached 80% of the maximal, 24-hour level in 4 hours. In addition, a 4-hour incubation of DU 145 cells on SU-ECM invasion substrates in the presence of an identical concentration of the scrambled HSPNR peptide control resulted in no detectable invasion. Thus, the time courses of PHSRN-induced invasion and PHSRN-induced MMP-1 secretion into the medium seem to correspond closely. This temporal correlation is consistent with the functioning of secreted MMP-1 in invasion by DU 145 cells.
Because active MMP-1 has been shown to associate with surface α2β1 integrin collagen receptors during cell migration (41), these results suggest that exposure to the PHSRN peptide should also increase levels of MMP-1 in cell lysates. As reported previously for breast cancer cells (9), PHSRN peptide treatment also increased the levels of MMP-1 found in DU 145 cell lysates as shown in Fig. 4C. In contrast, exposure to an equimolar concentration of the LDV peptide ligand of the α4β1 fibronectin receptor neither induced nor decreased MMP-1 expression, nor did it affect PHSRN-induced levels of MMP-1 despite its demonstrated role in regulating α5β1-mediated MMP-1 expression in α5β1+α4β1+ cells (42). These results are consistent with the known lack of α4β1 integrin fibronectin receptors on the surfaces of α5β1+ DU 145 cells (11).
Role of FAK, PI3K, and PKCδ in PHSRN-induced MMP-1 expression. The functional roles of FAK and PKCδ in PHSRN-induced MMP-1 expression in cell lysates were assessed by overexpressing FAK, PKCδ, or nonspecific siRNA in DU 145 cells. As indicated by the immunoblot shown in Fig. 5A, PHSRN peptide treatment increased the levels of MMP-1 associated with DU 145 cells and in DU 145 cells expressing nonspecific siRNA. However, no PHSRN-induced increase in MMP-1 was observed in DU 145 cells expressing either FAK or PKCδ siRNA. Figure 5B presents the results of multiple immunoblots similar to the example shown in Fig. 5A. This histogram depicts the mean levels of MMP-1 in lysates of DU 145 cells relative to the serum-free control. As in the example above, PHSRN-induced MMP-1 expression occurred only in DU 145 cells and in the DU 145 cells overexpressing nonspecific siRNA; no MMP-1 induction was observed in DU 145 cells overexpressing FAK or PKCδ siRNA. Thus, consistent with the results for PHSRN-induced invasion, both FAK and PKCδ activities also seem to be required for PHSRN-induced MMP-1 expression.
The requirements for PI3K and PKCδ activities in α5β1-mediated invasion also suggested that they would be required for PHSRN-induced MMP-1 secretion. Thus, serum-free, adherent DU 145 cells were treated with either the LY294002 PI3K or the Rottlerin PKCδ inhibitors; then, MMP-1 secretion was induced with the PHSRN peptide. Duplicate wells of cells were treated with Safingol (37) before PHSRN treatment. As shown in Fig. 5C, PHSRN-induced MMP-1 secretion was reduced by ∼6-fold, to background levels, by both PI3K and PKCδ inhibitors. However, the PKCα inhibitor had little effect on PHSRN-induced MMP-1 secretion. These results suggest that, consistent with their roles in PHSRN-induced invasion, PI3K and PKCδ also function in PHSRN-induced MMP-1 secretion.
We report that exposure of adherent, serum-starved DU 145 cells to the PHSRN peptide induces rapid phosphorylation of the FAK Y397 residue and FAK/PI3K p85 subunit association. Because FAK exhibits increased kinase activity and Y397 autophosphorylation after integrin activation (18), we evaluated the ability of the PHSRN peptide to induce PI3K activity. We found that PHSRN induces a burst of PI3K activity in DU 145 cells, which temporally corresponds to FAK Y397 phosphorylation and PI3K/FAK association. We tested the requirement for FAK in PHSRN-induced invasion by using the technique of RNAi (24) to down-regulate FAK expression specifically. We found that FAK siRNA treatment prevents PHSRN-induced DU 145 invasion, whereas nonspecific siRNA has no effect. We also found that, in addition to FAK, PI3K activity is required for PHSRN-induced invasion. Because the products of PI3K (PIP2 and PIP3) activate PKCδ directly (25), we assessed the role of PKCδ in PHSRN-induced invasion. By using siRNA and a specific inhibitor, we found that PKCδ also functions in PHSRN-induced invasion.
In addition, we found that, as observed previously for human breast cancer (9), MMP-1 activity is required for PHSRN-induced invasion by DU 145 cells, whereas gelatinases MMP-2 and MMP-9 do not seem to be involved. This is consistent with the abundant, native type I collagen found in SU-ECM (9). In addition, the time courses of PHSRN-induced invasion, FAK Y397 phosphorylation, FAK/PI3K p85 association, and MMP-1 secretion closely correspond. In addition, we found that PHSRN treatment up-regulates MMP-1 levels in the lysates of DU 145 cells as well as in their medium. Consistent with the roles of PI3K and PKCδ in α5β1-mediated DU 145 invasion, their specific inhibitors, LY 294002 and Rottlerin, prevent PHSRN-induced MMP-1 secretion. We also observed that FAK and PKCδ siRNA prevent the accumulation of MMP-1 in lysates of PHSRN-treated DU 145 cells, whereas nonspecific siRNA has no effect. Thus, our results show the requirement for FAK, PI3K, and PKCδ activities in PHSRN-induced MMP-1 secretion as well as for invasion. Consistent with our results, the involvement of PI3K in α5β1-mediated invasion has been observed for transformed mammary epithelial cells (10, 23). PI3K has also been shown to function in MMP-1-dependent invasion by Caco-2 colonic cancer epithelial cells (43).
These results suggest the intracellular signaling pathway for the induction of α5β1-mediated invasion shown in Fig. 6. PHSRN binding by α5β1, in the absence of the α4β1/pFn interaction, activates α5β1 to induce FAK phosphorylation at Y397 and promote the FAK/PI3K p85 interaction. This brings the catalytic p110 subunit to the membrane, where it phosphorylates PIP2 to form PIP3, which then activates PKCδ (25). Interestingly, PKCδ has been shown to play an important role in regulating transcriptional activation of proinflammatory nuclear factor-κB (NF-κB)–dependent genes, such as the VCAM-1 gene in endothelial cells (44). Moreover, NF-κB can also activate MMP-1 gene expression (45); thus, we speculate that PKCδ activation by PI3K-generated PIP3 may function to stimulate MMP-1 expression in prostate cancer cells.
The model shown in Fig. 6 is also consistent with the requirements for FAK and PI3K in migration on a fibronectin substratum. Many studies suggest that FAK promotes cell migration on fibronectin. Furthermore, expression of various mutants in FAK knockout cells suggests that both the kinase activity of FAK and its Y397 PI3K p85-binding site are individually required for cell migration to fibronectin. Moreover, FAK pY397 is necessary for the association of FAK with PI3K in vivo (18). It has also been shown that PI3K binding is required for FAK to promote cell migration on fibronectin, whereas Src binding is not sufficient. Moreover, inhibition of PI3K was found to decrease cell migration significantly (46).
Integrins are crucial to prostate cancer progression and metastasis, especially because they can induce cancer cell migration and invasion. Our previous research has focused on the importance of α5β1 in mediating the constitutive invasiveness of metastatic prostate and breast cancer cells. We have shown that α5β1 interacts with the PHSRN sequence of the fibronectin cell-binding domain (2, 6, 9) to induce invasion. Many lines of metastatic prostate and breast cancer cells are constitutively invasive because they lack cell surface α4β1 integrin (10, 11), which interacts with a distinct site on fibronectin to repress α5β1-mediated MMP-1 expression (42) and invasion (9). Several studies have also shown increased α5β1 integrin on cell surfaces in sectioned human and rat prostate cancer tumors relative to normal prostate tissue and observed that α5β1 integrin increases with loss of tumor cell differentiation, increased Gleason score, local progression, or metastasis (47, 48).
MMP-1 expression is a marker for tumor progression in many other types of cancer (49), and high levels of MMP-1 expression correlate with a poor prognosis (50). Moreover, increased MMP-1 expression in tumor cells is significantly correlated with the depth of tumor invasion, angiogenesis, lymphangiogenesis, and presence of local and distant metastases (51).
Sectioned primary tumors from radical prostatectomies exhibit reduced MMP-1 levels, consistent with the limited tissue destruction observed in many primary prostate tumors (52). In contrast, sections from prostatic bone marrow metastases exhibit strong immunostaining for MMP-1, and patient-derived, metastatic prostate cancer cells, cocultured with bone marrow stroma, also strongly stain for MMP-1. Based on these results, it was suggested that up-regulation of MMP-1 secretion may be an important factor in the formation of prostate cancer metastases (53). Consistent with the potentially causal relationship between MMP-1 expression and malignancy, MMP-1 expression is also significantly higher in the more aggressive prostate cancer cell lines and sublines (54). Interestingly, MMP-1 also cleaves entactin, thus contributing directly to the degradation of basement membrane and hence potentially to the transiting of epithelial barriers by tumor cells (55), in addition to stromal proteolysis. Thus, although many MMPs contribute to tumor angiogenesis and metastasis, MMP-1 may be critical for the development of the invasive phenotype during cancer progression. Our results suggest that the interaction of α5β1 integrin with the fibronectin PHSRN sequence and the resulting activation of FAK, PI3K, and PKCδ are very important steps in the intracellular signaling pathway leading to MMP-1-dependent invasion by metastatic human prostate cancer cells.
Note: Z-Z. Zeng and Y. Jia contributed equally to this work.
Current address for Y. Jia: Laboratory of Pathology, Center for Cancer Research, National Cancer Institute, NIH, Building 10, Room 2A29, Bethesda, MD 20892-1500.
Grant support: Attenuon L.L.C. (San Diego, CA) research grant (D.L. Livant) and University of Michigan summer research fellowship for medical students (N.J. Hahn).
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.
We thank Drs. Andrew Mazar and Graham Parry (Attenuon) for many helpful discussions and for peptide preparations, Dr. R.K. Brabec for assistance in the analyses of MMP-1 activities, and the laboratory of Dr. Theodore Lawrence for the use of their microplate spectrophotometer and their fluorescence microscope.