Glioblastoma multiforme are highly aggressive tumors for which no adequate treatment has yet been developed. Glioblastoma multiforme show large amounts of active Ras, considered an appropriate target for directed therapy. Here, we show that the Ras inhibitor S-trans, trans-farnesyl thiosalicylic acid (FTS) can avert the transformation of human glioblastoma multiforme cells by inhibiting both their migration and their anchorage-independent proliferation. FTS, by down-regulating Ras activity in glioblastoma multiforme cells, inhibited phosphatidylinositol 3-kinase signaling, resulting in decreased activity of Rac-1. At the same time, activation of RhoA was increased. These two small GTPases are known to control the arrangement of the actin cytoskeleton. By tilting the balance between Rac-1 and RhoA activities, FTS caused the glioblastoma multiforme cells to undergo profound changes in morphology, including rearrangement of actin into stress fibers and assembly of focal adhesions, both of which are governed by RhoA signaling. These morphologic changes allowed strong attachment of the cells to the matrix, rendering them immobile. The results show that FTS should be considered as a candidate drug for glioblastoma multiforme therapy because it targets not only cell proliferation but also cell migration and invasion, which together constitute the most problematic aspect of these malignancies. (Cancer Res 2006; 66(24): 11709-17)
- cell migration
Glioblastoma multiforme are the commonest of the primary brain tumors and among the most aggressive, with a median survival time from diagnosis of up to 1 year ( 1, 2). A prominent characteristic of these tumors is their ability to infiltrate healthy brain tissue, creating new malignant foci. This has thus far posed an unsolvable problem in attempts at therapy; total resection is almost never achieved, and patients who undergo surgery are subject to recurrences, usually in tissue adjacent to the resection area, consistent with the highly invasive nature of these tumors ( 3). Given this feature, it is obvious that efforts to battle this disease must be focused on its invasive nature. One way to solve the problem would be to effectively inhibit signaling pathways that control cell migration and invasion. Here, we examined whether S-trans, trans-farnesyl thiosalicylic acid (FTS), a specific inhibitor of Ras ( 4), can inhibit glioblastoma multiforme cell motility and transformation.
Ras is a small GTPase that mediates between extracellular signals and the intracellular environment and controls key cellular processes, such as proliferation and differentiation ( 5), cytoskeleton arrangement, cell migration ( 6, 7), and cell survival and death ( 8). When Ras is loaded with GTP, it activates an array of effectors, which in turn operate these cellular processes. Over the years, it has become clear that the activation of signaling pathways mediated by Ras does not occur in a simple linear fashion but rather that activating or inhibitory interactions between different cellular messengers create an intricate network ( 9). However, there are a few well-defined Ras-activated pathways that contribute to our understanding of key biological processes. Three classic Ras pathways are the Raf/mitogen-activated protein (MAP) kinase pathway that controls cell proliferation ( 10, 11), the phosphatidylinositol 3-kinase (PI3K)/Akt pathway that mediates cell survival and cytoskeleton arrangement ( 12, 13), and the Ral-GDS/Ral pathway that participates in cell proliferation and motility and in cytoskeleton arrangement ( 14, 15). In addition, Ras can regulate the activity of small G proteins that belong to the family of Rho GTPases ( 9), which also participate in cytoskeleton arrangement, cell cycle control, and gene expression ( 16).
Apart from its central role in controlling cellular signals, Ras is one of the most commonly mutated oncogenes in human cancers; >30% of all human tumors harbor a mutated Ras. Oncogenic mutations, particularly in positions 12, 13, or 61, lead to defective GTPase activity, creating constitutively active proteins ( 17). In the case of glioblastoma, although these malignancies do not harbor a constitutively active mutant Ras, they do express high levels of active Ras because of overexpression of growth factors and their receptors, which leads to Ras activation ( 18). We therefore suspected that Ras inhibition might prove to be effective in disrupting the malignant phenotype of glioblastoma. In earlier experiments done by our group, promoter and biochemical analyses in glioblastoma multiforme treated by FTS revealed that two important Ras-controlled transcription factors are down-regulated by the treatment. One is hypoxia-inducible factor-1, whose down-regulation leads to glycolysis shutdown ( 19); the other is E2F1, whose down-regulation is associated with cell cycle arrest ( 20).
The Ras inhibitor FTS was designed to mimic the farnesyl cysteine moiety of the COOH terminus of Ras. This modification was shown to be crucial for proper Ras anchorage to the plasma membrane. FTS, by competing with Ras for its binding sites, disrupts Ras anchorage to the plasma membrane, causing the protein to become dislodged to the cytosol, where it undergoes rapid degradation ( 4). We show here that FTS inhibits the motility of glioblastoma cells by down-regulating the PI3K pathway and shifting the balance between the activities of Rho- and Rac-GTPases. The result is a drastic change in cell morphology, which renders the cells incapable of proliferation and migration. These findings show that FTS, by interfering with the invasive nature of these cells, can be considered as a potential drug for glioblastoma therapy.
Materials and Methods
Cell culture procedures. Human glioblastoma U87 MG, U251 MG, and LN229 cells, kindly donated by Eric Holland (Memorial Sloan-Kettering Cancer Center, New York, NY), were grown in DMEM containing 10% FCS, 2 mmol/L l-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. The cells were incubated at 37°C in a humidified atmosphere of 95% air and 5% CO2.
Anchorage-independent growth in soft agar. Noble agar 2% (Difco, Detroit, MI) was mixed with medium (DMEM × 2, containing 20% FCS, 4 mmol l-glutamine, 200 units/mL penicillin, and 0.2 mg/mL streptomycin), and 100 μL of the mixture was poured into each well of 96-well plates to provide the base agar (at a final agar concentration of 1%). Agar (0.6%) was mixed with DMEM × 2 containing U87 or LN229 cells at a density that provided 8 × 104 per well, or U251 cells at a density that provided 4 × 104 per well, and 50 μL of the mixture was plated on the base agar (at a final concentration of 0.3%). FTS and Me2SO4 were prepared in DMEM × 1 at different concentrations, and 50 μL of mixture was placed in each well so that the final concentrations of FTS were 0, 25, 50, or 100 μmol/L, whereas the Me2SO4 concentration was adjusted to 0.1%. FTS, a gift from Concordia Pharmaceuticals (Sunrise, FL), was prepared as previously described in detail ( 21). The cells were incubated for 14 days and then stained for 4 hours with 1 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma-Aldrich, St. Louis, MO), which stains active mitochondria in living cells, and the colonies were imaged. Colonies larger than 0.08 mm2 (means ± SD, n = 18) were counted using Image-Pro Plus software (Media Cybernetics, Carlsbad, CA). The colonies in each treatment group were divided into three subgroups based on their size: 0.08 to 0.16, 0.16 to 0.32, and >0.32 mm2. The average percentage of colonies in each subgroup (means ± SD, n = 18) was calculated by dividing the number of colonies of a particular treatment and specific group size by the number of colonies of the same size in the corresponding untreated control group.
Scratch-induced migration assay. This was done as described elsewhere ( 22). In short, U87, U251, and LN229 cells were seeded on collagen-covered 35-mm plates at a density of 0.4 × 106 per plate. After 24 hours, the medium was replaced by 0.5% FCS containing DMEM, and the cells were treated for 24 hours with FTS (12.5 μmol/L) or with U0126 (25 μmol/L), LY294002 (10 μmol/L), or Y27632 (10 μmol/L). The last three inhibitors were from Calbiochem (La Jolla, CA). In each plate, three areas were scratched, creating three gaps of similar widths. The media and the inhibitors were then replenished. Immediately thereafter, and at the time points indicated in Results, phase-contrast images of the plates were obtained with a CCD camera connected to an Olympus fluorescent microscope (×10 objective). The region imaged at zero time was marked to enable us to photograph the same area at different times, thus allowing examination of a specific population of migrating cells. The widths of gaps treated with the different inhibitors and at different time points were measured by Image-Pro Plus software. The data acquired from the three scratches on each plate were averaged to obtain the mean gap width at a given time. Statistical analysis disclosed either the mean gap width (in arbitrary units) of FTS-treated cells relative to the control at different time points (means ± SD, n = 6) or the percentage of migration, calculated as the size of the gap that had closed at the final time point, expressed as a percentage of the gap size at zero time for each treatment (means ± SD, n = 6). For transfections, U87 cells were plated as detailed above and, 24 hours later, transfected either with green fluorescent protein (GFP) or with GFP-H-Ras17N using the FuGene reagent (Roche Molecular Biochemicals, Mannheim, Germany) according to the manufacturer's instructions. After 24 hours, the cell monolayer was scratched as detailed. Immediately thereafter, and 24 hours later, phase-contrast and fluorescent images of the plates were obtained (×10 objective). The gap width of the zero time point scratch was measured by Image-Pro Plus software. This area was then overlaid upon the 24 hours time point images, and the number of fluorescent cells was counted. Statistical analysis disclosed the ratio (means ± SD, n = 6) between the number of fluorescent cells that migrated into the gap and the total number of fluorescent cells in the image.
Western blot analysis. U87 cells were plated at a density of 0.8 × 106 or 1.7 × 106 cells in 10-cm or 14.5-cm dishes, respectively, and were allowed to grow overnight in medium containing 10% FCS. The medium was then replaced by medium containing 0.5% FCS, and the cells were treated for 24 hours with 0.1% Me2SO4 or 12.5 μmol/L FTS. We then lysed the cells with solubilization buffer [50 mmol/L Tris-HCl (pH 7.6), 20 mmol/L MgCl2, 200 mmol/L NaCl, 0.5% NP40, 1 mmol/L DTT, and protease inhibitors], and the lysate (20-100 μg) was subjected to SDS-PAGE and then immunoblotted with one of the following antibodies, as follows: anti-pan-Ras (1:2,500), anti-phospho-Akt (1:1,000), anti-Akt (1:2,000), anti-phospho-extracellular signal-regulated kinase (phospho-ERK; 1:10,000), anti-ERK2 (1:1,000), anti-phospho-S6K/Y389 (1:1,000), anti-β-tubulin (1:500), anti-RalA (1:2,000), anti-Rac-1 (1:2,500), anti-RhoA (1:700), anti-phospho-focal adhesion kinase (phospho-FAK; 1:1,000), and anti-FAK (1:1,000). The immunoblots were exposed either to 1:2,500 peroxidase-goat anti-mouse IgG or to 1:2,500 peroxidase-goat anti-rabbit IgG, and protein bands were visualized by enhanced chemiluminescence and quantified by densitometry with Image Master VDS-CL (Amersham Pharmacia Biotech, Arlington Heights, IL) using TINA 2.0 software (Ray Test). Mouse anti-pan-Ras antibody (Ab-3) was from Calbiochem; mouse anti-phospho-ERK antibody was from Sigma-Aldrich; rabbit anti-ERK antibody was from Santa Cruz Biotechnology (Santa Cruz, CA); rabbit anti-phospho-Akt antibody (Ser473; 4E2) was from Cell Signaling (Beverly, MA); rabbit anti-Akt antibody was from Cell Signaling; rabbit anti-phospho-S6K antibody (Tyr389) was from Sigma-Aldrich; mouse anti-β-tubulin antibody was from Sigma-Aldrich; mouse anti-RhoA antibody was from Upstate Biotechnology (Lake Placid, NY); mouse anti-RalA antibody was from Transduction Laboratories (Los Angeles, CA); rabbit anti-phospho-Akt antibody (Ser473; 4E2) was from Cell Signaling; rabbit anti-phospho-FAK antibody (pY397) was from BioSource (Camarillo, CA); rabbit anti-FAK antibody was from Santa Cruz Biotechnology; peroxidase-goat anti-mouse IgG and peroxidase-goat anti-rabbit IgG were from Jackson ImmunoResearch Laboratories (West Grove, PA).
GTPase pull-down assays. Lysates containing 1 mg protein were used to determine the Ras-GTP content by the glutathione S-transferase (GST)-RBD (Ras-binding domain of Raf) pull-down assay, as described elsewhere ( 23). Ral-GTP was pulled down by the use of lysates containing 1 mg protein and GST-RalBD (Ral-binding domain of RalBP1)–conjugated beads, as described ( 24). Rac-GTP and Rho-GTP were pulled down by the use of lysates containing 2 mg protein and GST-PBD (Rac-binding domain of PAK1)– and GST-Rhotekin BD (Rho-binding domain of Rhotekin)–conjugated beads ( 25, 26). In short, cells were lysed with lysis buffer [25 mmol HEPES (pH 7.5), 150 mmol NaCl, 1% NP40, 10% glycerol, 10 mmol MgCl2, 1 mmol/L EDTA, and protease inhibitors] and then incubated for 1 hour at 4°C with glutathione-Sepharose beads bound to 20 μg of GST-PBD or GST-rhotekin BD. The beads were then washed thrice in lysis buffer, resuspended, and boiled in sample buffer (×1), and the supernatants were subjected to SDS-PAGE followed by immunoblotting with the appropriate antibodies, as described above.
Immunofluorescence and transfections. U87 cells were seeded on glass coverslips in DMEM containing 10% FCS placed in six-well plates at a density of 1 × 105 per well for 24 hours. The medium was then replaced by medium containing 0.5% FCS, and 0.1% Me2SO4 or FTS (10 or 12.5 μmol/L) was added. After further incubation for 24 hours, the cells were fixed, permeabilized, incubated for 30 minutes with 200 μg/mL naive goat IgG (Jackson ImmunoResearch), and washed. The cells were incubated with mouse anti-vinculin antibody (1:400; Sigma-Aldrich) for 1 hour, and this was followed by extensive washing and incubation for a further 30 minutes with goat anti-mouse Cy2-conjugated antibody (1:200; Jackson ImmunoResearch) and phalloidin (1:1,000; Sigma-Aldrich). Slides were washed, then mounted, and imaged.
In transfection experiments, the cells were seeded as described above and, 24 hours later, were transfected with GFP-H-Ras17N or with an empty GFP vector, or cotransfected with GFP and Rac17N, or with GFP and pcDNA3 using the FuGene reagent (Roche Molecular Biochemicals) according to the manufacturer's instructions. After 24 hours, the cells were fixed, stained, and mounted as described except that vinculin staining was with Cy3-conjugated secondary antibody. Filamentous-actin (F-actin), focal adhesions, and GFP were visualized and then photographed under an LSM510 confocal microscope (×63 objective) fitted with fluorescein and rhodamine filters. For statistical analysis, cells with stress fibers and focal adhesions (in the case of transfection, only cells expressing GFP) were counted under an Olympus fluorescence microscope. Cells with stress fibers and focal adhesion (means ± SD, n = 6) were expressed as a percentage of 100 cells counted from each slide.
Ras inhibition blocks transformation of glioblastoma cells. Glioblastoma are notorious for their ability to recur within close proximity of the original tumor. This is because the highly transformed phenotype promotes rapid proliferation and migration ( 3). Although studies have shown that FTS can inhibit the anchorage-dependent growth of U87 cells ( 20), it was not known whether this Ras inhibitor can also reverse the transformed phenotype of glioblastoma. To examine this possibility, we first examined the effect of FTS on the anchorage-independent growth of U87 cells, using a soft agar assay in which U87 cells were treated with FTS (25, 50, or 100 μmol/L) or, as a control, with Me2SO4 (0.1%). After 14 days of treatment, the cells were stained with MTT and imaged. Colonies larger than 0.08 mm2 were counted. Results of a typical experiment ( Fig. 1A ) showed that U87 cells formed a significant number of colonies in the soft agar (222 ± 26.4, n = 18). FTS caused a dose-dependent decrease in colony number ( Fig. 1A). The decrease was highly significant (P < 0.05 at 25 μmol/L and P < 0.01 at 50 and 100 μmol/L FTS, n = 18; Fig. 1B). Notably, FTS also caused a decrease in the colony size, as evidenced by the much greater decrease in the number of large colonies than in the number of small colonies at any given concentration of FTS ( Fig. 1C). Taken together, these results indicated that FTS inhibits the anchorage-independent growth of U87 cells, although it is not unlikely that the decrease in total number of colonies also reflects cell death. We therefore concluded that FTS strongly attenuates the transformed phenotype of U87 cells. To examine whether this is a general effect of the Ras inhibitor on glioblastoma cells, we did a similar set of experiments with two additional glioblastoma cell lines U251 and LN229. The results of these experiments ( Fig. 1A and B) showed that FTS dose-dependently decreased colony formation of both cell lines in a manner similar to that observed in U87 cells.
Ras inhibition disrupts migration of glioblastoma cells. Next, we investigated the effect of FTS on a second manifestation of the transformed phenotype of glioblastoma (i.e., on cell motility). This was done by using a scratch-induced migration assay in which the width of the gap formed by a scratch was monitored at different times after the wound was inflicted ( 22). The glioblastoma cells were pretreated with FTS (12.5 μmol/L) or with Me2SO4 (0.1%) as a control, wounded 24 hours later, and maintained in 0.5% FCS-containing medium to block cell proliferation, which could otherwise account for gap closure. The gap width was then monitored at the indicated time points. Lower concentrations of FTS, compared with those used in the soft agar assays, were used in these and all subsequent experiments in which the cells were maintained at low serum. FTS is more effective in low serum than in high serum because it binds to serum proteins; thus, in low serum, the free drug concentration is higher than in high serum. Recent experiments showed that under these conditions, cell growth was almost completely halted ( 27). First, we determined the effects of the Ras inhibitor on active Ras at these low concentrations of serum and drugs. Treatment with FTS caused a time-dependent decrease in Ras-GTP ( Fig. 2C ). By 24 hours after treatment (i.e., at the time of wounding; and corresponding to zero time of the cell-migration assay), Ras-GTP levels in the FTS-treated cells were as low as 44 ± 4.4% of the levels in control cells. Thus, FTS effectively inhibited Ras under the conditions employed in the cell migration assays. Notable, Ras-GTP levels did not decrease in untreated cells within 24 hours ( Fig. 2C). Importantly, Hoechst staining experiments showed that within the time frame of these experiments FTS did not induce cell death (5–8% and 4–10% cell death, respectively, in control and FTS-treated cells) probably because of the relatively low FTS concentration used.
Next, we determined the effect of FTS on migration of U87 cells. Results of a typical experiment show that gap closure occurred within 18 to 24 hours in the control cultures ( Fig. 2A). Gap closure was strongly attenuated in the FTS-treated cells, which seemed to move at a slower pace than the control cells, closing smaller portions of the wound gap. By the time the 18-hour time point was reached, the FTS-treated cells had become almost totally static. Statistical analysis of the results indicated that FTS had caused a robust decrease in U87 cell migration ( Fig. 2B). The effect of FTS on cell migration was mimicked by expression of the dominant-negative mutant H-Ras(17N), which inhibited U87 cell migration (40 ± 2.5% inhibition, n = 6; Fig. 2D). Thus, both pharmacologic inhibitions of Ras and dominant-negative Ras inhibited U87 cell migration. Separate experiments showed FTS inhibition of cell migration was not limited to U87 cells as migration of U251 and LN229 cells was also significantly inhibited by FTS ( Fig. 2E).
FTS induces inhibition of Ras signaling in U87 cells under conditions of low serum concentration. The results described above showed that under conditions in which FTS strongly inhibits Ras, it also exhibited a strong inhibitory effect on the migration of U87 cells. We therefore next examined the effects of FTS on the activity of effectors that operate downstream of Ras and are known to participate in regulation of cell migration. First, we examined the effects of FTS on the levels of phospho-Akt, phospho-ERK, and Ral-GTP as readouts of the prominent Ras pathways PI3K, Raf/ERK kinase (MEK)/ERK, and RalGDS, respectively ( 28). Results of typical experiments are shown in Fig. 3A and B , and quantitative analysis is given in Fig. 3C. Relative to control, FTS (12.5 μmol/L, 24-hour treatment) caused a significant decrease in phospho-Akt (by 34 ± 7%, n = 4; P < 0.05) and in phospho-S6-kinase (by 39 ± 15.3%, n = 4; P < 0.05) a downstream effector of Akt but did not affect the level of phospho-ERK. FTS did not induce a change in the total levels of Akt and ERK ( Fig. 3A). FTS caused, however, a significant reduction in the level of Ral-GTP (by 38 ± 12%, n = 4; P < 0.05; Fig. 3B). The decreases in the levels of phospho-Akt and phospho-S6-kinase are consistent with recent observations that FTS inhibited PI3K activity in U87 cells ( 19). It thus seems that under the conditions in which we observed strong inhibition of U87 cell migration and a reduction in Ras-GTP ( Fig. 2), the pathways affected were the PI3K and RalGDS, but not the Raf/MEK/ERK.
Next, we examined the effects of FTS on signaling molecules that are controlled by the Ras/PI3K and Ras/Ral pathways. Figure 3B and C show that, relative to control, FTS caused a significant reduction in Rac-GTP (by 55 ± 11.2%, n = 8; P < 0.01), whose activation is controlled by several Ras-activated molecules, including Tiam 1 ( 29), PI3K ( 12), and SOS ( 30). It is worth noting that Rac activation is inhibited by the Ral effector RalBP1, which acts as a Rac GTPase-activating protein ( 31). It therefore seems that the observed FTS-induced reduction in Ral-GTP might not account for the reduction in Rac-GTP, as such an effect would have the opposite outcome (i.e., an increase in Rac-GTP). Interestingly, FTS caused a marked increase in RhoA-GTP (by 2.33 ± 0.36-fold, n = 6; P < 0.05; Fig. 3B and C). This might be a reflection of the antagonistic relationship known to exist between the activation of Rac and of Rho ( 32– 34). Whether or not this is the case, the observed FTS-induced increase in RhoA-GTP points to the apparent selectivity of FTS for the farnesylated Ras; lack of selectivity would have resulted in decrease in the active prenylated Rho protein as well.
PI3K pathway is essential for U87 cell migration. The above findings led us to study the contributions of PI3K-dependent and Rho-dependent pathways to the FTS-induced inhibition of U87 cell migration. Accordingly, we examined how U87 cell migration is affected by the specific PI3K inhibitor LY294002 and by the specific inhibitor Y27632 of the Rho effector Rho-kinase/ROCK. We also examined the effect of the specific MEK inhibitor U0126 on U87 cell migration. Recent experiments showed that the inhibitors effectively inhibit their corresponding targets in U87 cells ( 27). The results of these experiments are summarized in Fig. 4 . Relative to control, LY294002 (10 μmol/L) caused a significant decrease in migration of U87 cells (by 34%, n = 12; P < 0.05), although not to the same extent as the FTS-induced inhibition of cell migration (by 60%, n = 12; P < 0.01). Neither the ROCK inhibitor Y27632 (10 μmol/L) nor the MEK inhibitor U0126 (25 μmol/L) inhibited U87 cell migration ( Fig. 4). These results suggest that signaling by the Ras/PI3K pathway, but not by the Ras/ERK pathway, plays an important role in migration of U87 cells. This conclusion is consistent with the observed inhibition of phospho-Akt but not of phospho-ERK by FTS ( Fig. 3), suggesting that the inhibitory effects of FTS on cell migration are associated with inhibition of the PI3K pathway. The lack of inhibition of U87 cell migration by Y27632 is not unexpected in light of the observed FTS-induced activation of Rho ( Fig. 3). This is because FTS and Y27632 have opposite effects on the Rho/ROCK pathway: it is activated by FTS ( Fig. 3) but inhibited by Y27632 ( 35).
FTS causes a drastic morphologic change in U87 cells accompanied by reorganization of the actin cytoskeleton. The results described above led us to examine the effects of FTS on U87 cell morphology and cytoskeleton organization. This is because previous studies have shown that PI3K mediates the activation of Rac, which in turn participates in the control of cell migration by regulating the actin cytoskeleton ( 36). We therefore examined the effect of FTS on actin and on vinculin, using rhodamine-labeled phalloidin, which interacts with polymeric F-actin, and an anti-vinculin antibody to label vinculin, which is recruited to focal adhesions at their assembly ( 37). The cells were treated with FTS (10 or 12.5 μmol/L) or with the vehicle control (0.1% Me2SO4) for 24 hours under the same conditions as those used in the scratch-induced migration assay and then imaged. Typical phase-contrast images of control and FTS-treated cells illustrate the marked change in morphology that was induced by the Ras inhibitor. Whereas control cells showed small cell bodies and long extensions, the FTS-treated cells exhibited flat morphology and were much larger ( Fig. 5A ). Fluorescent images of the stained cells (red, polymeric actin; green, vinculin) are shown in Fig. 5B. Most of the polymeric actin in the control U87 cells seemed to be concentrated in membrane ruffles, and the vinculin seemed to be diffused in the cell. In the FTS-treated cells, on the other hand, polymeric actin became organized in stress fibers and the membrane ruffles disappeared ( Fig. 5B), whereas strong vinculin staining was observed mainly in focal adhesions located at the edges of stress fibers (see merged images in Fig. 5B). Cell counts disclosed that stress fibers and focal adhesions were exhibited by 63 ± 5.9% of the FTS-treated cells compared with only 27.5 ± 4.3% of the control cells ( Fig. 5C). Separate experiments showed that dominant-negative Ras (Ras17N) mimicked the effects of FTS on the actin cytoskeleton and focal adhesions ( Fig. 5B and C), confirming the specificity of the inhibitor for the active Ras protein. Dominant-negative Rac (Rac17N) also mimicked the effects of FTS ( Fig. 5C). These results are in line with the observed FTS-induced up-regulation of Rho-GTP and down-regulation of Rac-GTP ( Fig. 3), which respectively control the formation of stress fibers and membrane ruffles ( 15). The FTS-induced increase in stress fibers and focal adhesions suggested that their disassembly had been attenuated, rendering the cells immobile. We therefore examined the effect of FTS on activation of FAK, which controls the turnover of focal adhesions ( 38). Figure 5E shows that FTS down-regulated the active autophosphorylated form of FAK (phospho-FAKY397).
FTS-induced modulation of the actin cytoskeleton in U87 cells is mimicked by a PI3K inhibitor. Taken together, the above results strongly suggested that FTS mediates inhibition of U87 cell migration by modulating the PI3K/Rac- and Rho-controlled actin cytoskeleton organization and turnover of focal adhesions. We therefore next examined the effect of PI3K inhibition on the actin cytoskeleton in U87 cells. The cells were treated with LY294002 (10 μmol/L) for 24 hours and then stained for polymeric actin and vinculin as described above. This treatment caused the disappearance of membrane ruffles and the appearance of stress fibers and focal adhesions almost indistinguishable from those induced by FTS ( Fig. 6A ). Statistical analysis indicated that treatment with LY294002 more than doubled the percentage of cells exhibiting stress fibers and focal adhesions (from 28% to 62%; Fig. 6B). Thus, the PI3K inhibitor LY294002 clearly mimicked the effect of FTS. In marked contrast the MEK inhibitor U0126 (25 μmol/L), which did not inhibit U87 cell migration, had no effect on actin or vinculin organization ( Fig. 6B). Interestingly, the ROCK inhibitor Y27632 (10 μmol/L) caused a small but significant decrease in the percentage of cells exhibiting actin stress fibers and focal adhesions (from 28% to 16%, Fig. 6B). The effect could be expected as this inhibitor has the opposite effect to that of FTS on the Rho/ROCK-dependent pathway.
Most glioblastoma harbor a chronically active Ras ( 18), whose immense contribution to malignant phenotypes has been shown to involve activation of signaling pathways that control proliferation, survival, and cell migration ( 5– 8). In the present study, we showed that the Ras inhibitor FTS reversed the malignant phenotype of U87, U251, and LN229 glioblastoma cells by inhibiting both cell migration ( Fig. 2) and anchorage-independent growth ( Fig. 1). The inhibition was accompanied by extreme alterations in cell morphology, attributed to reorganization of the actin cytoskeleton ( Fig. 5). FTS caused the U87 cells to become larger and flatter, and the polymeric actin, which is concentrated mainly in membrane ruffles in the untreated U87 cells, was reorganized into stress fibers. Moreover, focal adhesions were apparent only after FTS treatment and were localized to the extremities of the actin fibers ( Fig. 5). The significance of the morphologic changes lies in their implications for cell motility ( 36): once the actin is organized into stress fibers, and focal adhesions are assembled, the cells flatten and become attached to the extracellular matrix. Cells in motion need to assemble and disassemble actin structures to progress by alternating between attachment to and detachment from the ECM ( 36). Treatment with FTS caused an increase of >2-fold in the percentage of cells exhibiting flattened morphology with strong stress fibers and focal adhesions ( Fig. 5), suggesting that the disassembly of stress fibers and focal adhesions was attenuated, rendering the cells immobile. These findings are highly important when contemplating novel treatments for glioblastoma. This is because the main problem with these malignancies is their tendency to diffuse, conveyed by single cells leaving the core lesion and invading healthy brain tissue, and eventually creating recurrent tumors that harbor greater resistance to common treatments ( 3). Application of FTS, which might not only slow tumor growth but also decrease the invasiveness of the tumor cells, would bring about progress in glioblastoma therapy.
Our results strongly suggest that of the three prominent Ras pathways Raf/MEK/ERK, RalGDS/Ral, and PI3K/Akt, it is mostly if not only the last that participates in the FTS-induced inhibition of glioblastoma multiforme migration. Several observations support this conclusion: (a) The FTS-induced inhibition of U87 cell migration was mimicked by direct inhibition of PI3K by the inhibitor LY294002 but not by the MEK inhibitor U0126 ( Fig. 4); (b) LY294002 also mimicked the effect of FTS on U87 cell morphology ( Fig. 6); (c) FTS caused a significant decrease in phospho-Akt, the downstream effector of PI3K ( Fig. 3); (d) Ral activation by RalGDS-CAAX, found by us to enhance stress fiber formation in U87 cells, 1 was clearly inhibited by FTS ( Fig. 3), ruling out the involvement of active Ral in the FTS-induced alterations of U87 cell morphology.
The PI3K/Akt signaling pathway is known to promote cell cycle progression and survival, enabling tumor cells to thrive ( 13). PI3K can also promote the activity of Rac-1, which regulates growth factor–induced membrane ruffling ( 36). In the present study, FTS treatment caused a significant decrease in Rac-1 activity, while at the same time, it increased the activity of RhoA ( Fig. 3), which induces formation of stress fibers and focal adhesions ( 16). Together, these two GTPases enable the dynamic process of detachment of cells from the ECM and their reattachment to it and therefore are essential for cell migration ( 36). Studies have shown that activation of one of these proteins can abolish the actin organization governed by the other, suggestive of a morphologic antagonism between Rac-1 and RhoA ( 32– 34). FTS, by tilting the balance between these two GTPases in favor of RhoA, causes the actin organization governed by RhoA to become dominant. Consistent with this notion, dominant-negative Rac, like FTS, induced formation of stress fibers in U87 cells ( Fig. 5). These results are in contrast with the observation that dominant-negative Rac-inhibited stress fiber formation and cell migration in C6 rat glioma cells ( 39). The difference may be attributed to species or tumor cell type differences.
As illustrated in a simple model ( Fig. 6B), the combined results show that in U87 cells, the Ras inhibitor FTS causes down-regulation of PI3K, which in turn results in the inhibition of Rac-1 activity and an increase in RhoA activity. Consequently, the actin is rearranged into stable stress fibers, and focal adhesions are stabilized as well. The cells become attached to the matrix and remain immobilized. This might also explain the mechanism by which FTS causes strong inhibition of U87 anchorage-dependent and anchorage-independent growth. If FTS causes the cells to “program” themselves for attachment, then cell division, which requires detachment of cells from the matrix, will not take place, and proliferation will not be possible. In an anchorage-free environment, attachment will not be allowed. This might cause cell stress, thus preventing proliferation and perhaps even promoting cell death. Whether or not this is indeed the way in which FTS-induced inhibition operates, it is clear that FTS is highly potent in terms of inhibiting cell growth and migration of glioblastoma.
It is worth noting that the outcome of Ras inhibition on Ras pathways and on cellular behavior are strongly dependent on the tumor cell type ( 8, 40) and are also likely to be influenced by the cancer cell microenvironment ( 41). The results of our experiments with glioblastoma multiforme and other tumor cell lines further supports this notion. Thus, in an enriched serum environment that provides growth and survival factors, both the Raf/MEK/ERK and the PI3K pathways are highly active and are strongly inhibited by FTS ( 19, 28, 42). Such conditions allow inhibition of cell growth and induction of cell death, albeit at relatively high FTS concentrations. The high concentrations are needed because FTS is strongly adsorbed to the serum proteins, 1 leaving relatively low concentration of free drug. By employing specific conditions (low serum and relatively short periods of incubation) in the present experiments, we were able to show that the inhibition of U87 cell migration by FTS was achieved mainly because the PI3K/Akt and Rac/Rho pathways were affected ( Fig. 3). We could also rule out the possibility that FTS affects U87 cell migration through inhibition of the Raf/MEK/ERK pathway ( Fig. 3). Together, these experiments also highlight the importance of in vivo environmental conditions that are likely to determine the response to FTS. These include the availability of growth and survival factors in the vicinity of the tumor cells and the accessibility of the drug. The latter is certainly of particular importance in the case of brain tumors. In conclusion, our experiments suggest that application of FTS or one of its derivatives might not only slow the growth of glioblastoma multiforme but also decrease their invasiveness.
As pointed out earlier, an ongoing problem in glioblastoma therapy is drug accessibility. Many chemotherapeutic agents do not reach these tumors in adequate concentrations because of the blood−brain barrier (BBB). Recent innovations in this field have, therefore, been focused on the development of agents that will cross the BBB in sufficient concentrations, or on attempting to circumvent the BBB by mechanically disrupting it and introducing chemotherapeutic agents intra-arterially or by intra-cavity routes (using chemotherapy wafers or convection-enhanced delivery; ref. 43). None of these approaches, however, has yet to be shown to improve patient outcome to any significant extent. FTS was shown to penetrate the brain, although at concentrations in the micromolar range ( 44), and it is not yet known if such relatively low concentrations would be sufficient to induce glioblastoma multiforme cell death. The results described here point to the possibility that glioblastoma multiforme cell migration is likely to be inhibited in vivo by FTS. In our attempts to increase the accessibility of FTS-like Ras inhibitors, we have now prepared new FTS derivatives that exhibit significant potency in inhibiting U87 growth. 1 These compounds were designed to increase the hydrophobic nature of FTS, thus allowing better penetration of the brain.
Grant support: Israel Science Foundation grant 339/02 and Wolfson Foundation Trust.
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 S.R. Smith for editorial assistance.
Note: Yoel Kloog is the incumbent of The Jack H. Skirball Chair in Applied Neurobiology.
↵1 Unpublished observations.
- Received May 23, 2006.
- Revision received August 30, 2006.
- Accepted October 9, 2006.
- ©2006 American Association for Cancer Research.