Suppression of survivin expression in glioblastoma cells by the Ras inhibitor farnesylthiosalicylic acid promotes caspase-dependent apoptosis

Active Ras and phosphatidylinositol 3-kinase (PI3K)–dependent pathways, present in abundance in most glioblastoma multiformes (1, 2), contribute to maintenance of the malignant phenotype of these brain tumors by disrupting glioblastoma multiforme cell cycle arrest (3), increasing cell migration (4), enhancing glioblastoma multiforme cell survival (5), and promoting angiogenesis (6). The increased presence of active Ras-GTP in glioblastoma is secondary to mitogenic signals originating from activated receptor tyrosine kinases (7). Receptor-mediated activation of Ras signaling thus seems to be required for maintenance of the malignant phenotype of glioblastoma multiforme. Pharmacologic inhibitors of receptor tyrosine kinases (8) and Ras (9–11) or expression of dominant-negative Ras (2) inhibit glioblastoma multiforme cell growth, mainly by inhibiting Ras signaling to Raf/mitogen-activated protein/extracellular signal-regulated kinase (ERK) kinase (MEK)/ERK and to PI3K/Akt.

Consistently with the complex Ras-associated phenotype of glioblastoma multiforme, we found that the Ras inhibitor farnesylthiosalicylic acid (FTS) inhibits U87 glioblastoma multiforme cell growth, induces cell death, and down-regulates hypoxia-inducible factor-1α (HIF-1α; ref. 9). These changes in the malignant phenotype were associated with inhibition of active Ras (K-Ras, N-Ras, and H-Ras isoforms) and attenuation of Ras signaling to ERK, PI3K, and Akt (9). By applying computational methods, in which clustering of gene expression profiles was combined with promoter sequence analysis to obtain global dissection of the transcriptional response to FTS, we found that the transcription factor E2F1 was the predominant Ras-dependent component associated with cell cycle arrest (10). Evidently, inhibition of Ras by FTS promoted proteasomal degradation of cyclin D1, with a concomitant decrease in phosphorylated retinoblastoma protein accompanied by down-regulation of E2F1 and decreased expression of key E2F1-regulated genes critical for cell cycle progression. U87 cell growth arrest induced by FTS was overridden by constitutive expression of E2F1 (10).

A related study with U87 and other glioblastoma multiforme cell lines showed that FTS causes disappearance of HIF-1α, which is followed by reduced expression of key glycolysis pathway enzymes and other HIF-1α-regulated genes (9). These changes lead to glycolysis shutdown and a dramatic reduction in ATP levels, resulting in a severe energy crisis (9). It thus seemed that the abovementioned FTS-induced death of U87 cells (9) might be attributable, at least in part, to loss of energy. Other experiments suggested, however, that cell death induced by FTS invokes additional mechanisms that may or may not be associated with energy loss. It was shown recently, for example, that down-regulation of HIF-1α by phosphorothioate antisense HIF-1α oligonucleotide causes apoptotic cell death in U87 cells (12). That study showed that the cytotoxic effect of the antisense HIF-1α is independent of p53 and can be attenuated by caspase inhibitors. In addition, a recent study showed that FTS induces caspase activation and apoptosis in glioblastoma cells (13). Interestingly, a recent study showed that inhibition of K-Ras expression in a K-Ras glioblastoma mouse model results in apoptotic tumor regression (14). However, the studies described above and our own experiments did not find the link between Ras inhibition in glioblastoma multiforme and the consequent activation of cell death programs.

In the present work, we show that a major effect of the blockage of Ras pathways in glioblastoma multiforme cells by FTS is the decrease in survivin gene transcripts. This effect was also apparent at the level of survivin protein, which was reduced to an almost undetectable amount by FTS or by expression of dominant-negative Ras. Consequently, U87 cells underwent caspase-dependent apoptotic cell death. The cells could be rescued from FTS-induced death by forced expression of survivin.

The glioblastoma cell lines used and their growth conditions [U87, U87-E2F1, DBTRG-05MG (20/20), LN229, and U373] have been described previously (10). U87 cells stably expressing survivin (U87-survivin cells) were established by transfection with pcDNA3-HA-survivin expression vector followed by G418 (1 mg/mL) selection. FTS was prepared as described (15). Cells were plated at a density of 2.5 × 10⁵ in six-well plates for immunofluorescence assays and 1.5 × 10⁶ in 10-cm dishes for all other assays. Drugs or vehicle (0.1% DMSO) was added to the cells 24 hours after plating. FTS, U0126, or LY294002 was added at the concentrations indicated. Hoechst dye 33258 was obtained from Sigma-Aldrich (St. Louis, MO); the enhanced chemiluminescence kit was from Amersham (Arlington Heights, IL); mouse anti-pan-Ras (antibody-3) and mouse anti-p53 (antibody-1) antibodies were from Calbiochem (La Jolla, CA); mouse anti-β-tubulin (AK-15) and mouse anti-phosphorylated ERK antibodies were from Sigma-Aldrich; rabbit anti-survivin (FL-142), rabbit anti-E2F1 (C-20), monoclonal mouse anti-E2F1 (KH95), and rabbit anti-caspase-3 (H-277) antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA); rabbit anti-phosphorylated Akt (Ser⁴⁷³; 4E2), rabbit anti-Bad, rabbit anti-phosphorylated Bad (Ser¹³⁶), rabbit anti-caspase-7, and mouse anti-caspase-9 (C9) antibodies were from Cell Signaling (Beverly, MA); mouse anti-β-catenin antibody was from BD Transduction Laboratories (San Jose, CA); mouse anti–cytochrome c (556433) was from BD PharMingen (San Diego, CA); rabbit anti-Bax NH₂-terminal antibody was from Upstate Biotechnology (New York, NY); mouse monoclonal anti-poly(ADP-ribose) polymerase antibody (SA-250) was from Biomol Research Laboratories (Plymouth Meeting, PA); mouse anti-HA-tag antibody (hybridoma supernatant) was a gift from Y. Henis (Tel Aviv University, Tel Aviv, Israel); peroxidase-goat anti-mouse IgG, peroxidase-goat anti-rabbit IgG, and peroxidase-donkey anti-goat IgG antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA); LY294002 and U0126 were from Calbiochem; zVAD (Z-VAD-FMK) and Z-DEVD-FMK were from ICN Biomedicals (Aurora, OH); and QVD (Q-VD-OPH) was from MP Biomedicals (Aurora, OH).

For cell proliferation assays, we plated 1.5 × 10³ cells in 96-well plates or 3.5 × 10³ cells in 24-well plates. The cells were treated, 24 hours after plating, with different concentrations of FTS or with the vehicle (0.1% DMSO). Cell proliferation was determined after 3 days by using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay, which determines mitochondrial activity in living cells. The cells were incubated with 0.1 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide for 1 hour at 37°C and then lysed with 100% DMSO. Results were quantified by recording the absorbance at 570 to 630 nm. For fluorescence-activated cell sorting analysis, we plated 1 × 10⁶ cells in 10-cm dishes and treated the cells with 75 μmol/L FTS 24 hours later. Control cells were treated with vehicle only. After 48 hours, the cells were collected and suspended in PBS containing propidium iodide (50 μg/mL; Sigma-Aldrich) and 0.05% Triton X-100 (BDH, Poole, United Kingdom) for nuclear staining and then analyzed with a fluorescence-activated cell sorter (FACSCalibur, Becton Dickinson, Franklin Lakes, NJ).

All assays were done in triplicate or quadruplicate. Data were expressed as mean values ± SD. Statistical significance was determined by an unpaired Student's t test.

To examine the effects of inhibitors on survivin, phosphorylated ERK, phosphorylated Akt, E2F1, p53, and β-catenin, we used lysates of drug-treated or control (vehicle treated) cells and did Western immunoblotting analysis or Ras-GTP assay (16). Ras-GTP was determined by the glutathione S-transferase–Ras-binding domain of Raf pull-down assay as described previously (16). Protein bands were visualized by enhanced chemiluminescence and quantified as described previously (16). All biochemical and immunoblotting assays were repeated at least thrice. Data are presented as mean values ± SD.

U87 glioblastoma cells were cultured on glass coverslips and incubated with FTS or vehicle as described above. Alternatively, the cells were transfected with vectors (total amount, 2.3 μg DNA) encoding green fluorescent protein (GFP), GFP-H-Ras(17N), pcDNA3-E2F1, or pcDNA3 (10) using the JetPEI reagent according to the manufacturer's instructions (Talron, Rehovot, Israel). The cells were fixed, permeabilized, and labeled 48 hours after drug treatment or transfection with anti-survivin antibody (5 μg/mL) and Cy3-conjugated donkey anti-rabbit secondary antibody (2 μg/mL), with anti–cytochrome c antibody (5 μg/mL) and Cy3-conjugated donkey anti-mouse secondary antibody (2 μg/mL), or with rabbit anti-Bax NH₂-terminal antibody and Cy3-conjugated donkey anti-rabbit secondary antibody (2 μg/mL) as described elsewhere (9). The cells were also labeled with Hoechst dye 33258 as described earlier (9). Fluorescence images of survivin, Bax, or cytochrome c (Cy3, red), cell nuclei (Hoechst, blue), and GFP or GFP-H-Ras(17N) (green) were collected using a fluorescence microscope (IX70, Olympus America, Melville, NY) as described previously (9).

The expression vectors pRSV-GFP-H-Ras(17N) and pcDNA3-E2F1 have been described previously (10). pcDNA3-HA-survivin plasmid was a gift from D.C. Altieri (University of Massachusetts Medical School, Worcester, MA).

Extracts of total RNA (1 μg) from cells treated for 48 hours with FTS (70 μmol/L) or vehicle (control) were reverse transcribed in a total volume of 20 μL using the iSCRIPT cDNA kit (Bio-Rad Laboratories, Hercules, CA). The cDNA samples were used for real-time PCR (SYBR Green PCR kit, Roche Diagnostics Mannheim, Germany). For the survivin gene, we used 5′-ACCACCGCATCTCTAC3′ (forward primer) and 5′-TCCTCTATGGGGTCGT-3′ (reverse primer); for the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase, we used 5′-CCAGAACATCATCCCTGC-3′ (forward primer) and 5′-GGAAGGCCATGCCAGTGAGC-3′ (reverse primer). PCR amplifications were carried out using the LightCycler version 3.5 (Roche Diagnostics, Basel, Switzerland) according to the manufacturer's instructions. cDNA samples of control and FTS-treated cells (at least two dilutions for each) and cDNA standards were coamplified in the same reaction plate. The standard curve showed a linear relationship between the cycle threshold (C_T) values and the cDNA concentration. The relative expression of each gene was normalized using the expression levels of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase as standard.

A previous study by our group showed that the Ras inhibitor FTS induces marked changes in gene expression, as evident from DNA microarray analysis of cell samples tested at zero time and after drug treatment for 24 and 48 hours or of vehicle-treated controls (9). In the earlier experiments, we focused on two groups of genes (i.e., HIF-1α-regulated and E2F1-regulated genes) that were down-regulated by the Ras inhibitor (9, 10). In the present study, we focused on a third group whose expression was altered by the FTS treatment (i.e., a group of genes that are associated with cell death mechanisms; Fig. 1A). Heat-map image of the 30 death-associated genes that were up-regulated and of the 23 death-associated genes that were down-regulated by the FTS treatment is shown in Fig. 1A. Within this group, the gene whose expression was most dramatically altered was survivin (BIRC5). At 24 and 48 hours of FTS treatment, survivin expression was found to be decreased by 2.3-fold (log ratio, −1.2) and 78.8-fold (log ratio, −6.3), respectively (Fig. 1A and B). In addition, there were alterations in the expression of transcription factors that control survivin transcription (17–19), including E2F1, p53, β-catenin, and TCF4 (Fig. 1B). Although we found additional transcriptional changes in death-associated genes (e.g., in Bax, caspase-1, caspase-4, caspase-5, SKIP3, TRADD, and TRAF1; Fig. 1A), the dramatic change observed in survivin expression and the changes in its regulators led us to suspect that the observed FTS-induced glioblastoma cell death (9) might involve mechanisms associated with survivin. Real-time PCR analysis validated the gene expression results and showed that FTS (75 μmol/L; 48 hours of treatment) induced a decrease of 26 ± 0.006–fold in the expression of survivin (Fig. 1C).

Next, we examined whether the FTS-induced decrease in survivin expression was also manifested at the protein level. The effect of FTS on the amounts of survivin in U87 cells was determined by Western immunoblotting with anti-survivin antibody. The cells were treated with 75 μmol/L FTS and survivin was assayed at various times after the treatment (see Materials and Methods). To confirm anti-Ras activity, we assayed active Ras-GTP in the same samples. In agreement with earlier studies (9, 10), we found that FTS induced a time-dependent decrease in Ras-GTP (Fig. 1D); concomitant with this decrease, we found a time-dependent decrease in survivin protein (Fig. 1D). Importantly, in accord with the observed decrease in survivin mRNA (Fig. 1C), a 75% decrease in survivin protein was observed at 36 hours of FTS treatment, a time point at which the recorded sub-G₁ population of the cells was only 11% of the total cell population (9). This indicated that the decrease in survivin precedes death of the cells. Survivin levels continued to decrease, becoming almost undetectable at 48 hours of FTS treatment (Fig. 1D).

We next determined the effect of FTS on the amounts of survivin in three additional glioblastoma multiforme cell lines: DBTRG-05MG (20/20), LN229, and U373. The cells were incubated with 75 μmol/L FTS for 48 hours and then lysed and subjected to Western immunoblot analysis with anti-survivin antibody. Treatment with FTS resulted in a dramatic decrease in survivin in all cell lines, with the amounts of survivin recorded in 20/20, LN229, and U373 cells corresponding respectively to 15%, 2%, and 4% of that found in the untreated control cells (Fig. 2A). These results indicated that the effect of FTS on survivin is not restricted to U87 cells. Interestingly, U87 and 20/20 cells expressed wild-type p53, whereas LN229 and U373 cells expressed mutant p53. It therefore seemed that the FTS-induced down-regulation of survivin in glioblastoma multiforme might proceed by p53-independent mechanisms.

We next examined whether the FTS-induced decrease in survivin was a consequence of FTS-induced growth arrest or of inhibition of Ras-dependent pathways that control survivin expression (20). We did this by attenuating the growth of U87 cells by lowering serum concentrations and assaying survivin in the growth-attenuated cells. Under the conditions used in these experiments, no cell death was observed. Growth of U87 cells in 1% serum was, however, reduced by ∼2-fold relative to their growth in 20% serum (Fig. 2B). Despite this reduction in the rate of cell proliferation, the survivin levels remained relatively high (Fig. 2B), suggesting that survivin expression in U87 cells may not be dependent on the rate of U87 cell growth. Thus, FTS-induced inhibition of cell growth (10) is evidently not more than partially responsible, if at all, for the drug-induced decrease in survivin (Fig. 1C). This is consistent with earlier studies, which showed that survivin expression induced by active H-Ras is independent of cellular proliferation (20).

We next examined the effect of Ras inhibition on survivin by using an independent approach to inhibit Ras [i.e., by the use of GFP-tagged H-Ras(17N), a dominant-negative Ras isoform]. U87 cells were transfected with GFP-H-Ras(17N) or with GFP vector as a control. The cells were then stained with rabbit anti-survivin antibody and then with Cy3-conjugated anti-rabbit antibody. Typical dual fluorescence images (shown in Fig. 2C) show that, compared with the levels of survivin (red, Cy3 fluorescence) in both the cytoplasm and the nucleus of the control GFP-expressing cells (green fluorescence), the corresponding levels in the cells expressing GFP-H-Ras(17N) were much lower. GFP-H-Ras(17N) transfectants with no nuclear survivin accounted for as much as 80% of the total number of GFP-H-Ras(17N)-expressing cells compared with only 20% for the control GFP transfectants (Fig. 2C). Taken together, these experiments showed that inhibition of Ras by two independent methods [i.e., FTS treatment and GFP-H-Ras(17N) expression] leads to down-regulation of survivin in glioblastoma multiforme cells.

Because both the Ras/Raf/MEK/ERK pathway and the Ras/PI3K/Akt/glycogen synthase kinase-3 pathway participate in the regulation of survivin expression (20), we next examined the effects of inhibitors of these pathways on the levels of survivin in U87 cells. The cells were incubated for 24 hours with the MEK inhibitor U0126 or with the PI3K inhibitor LY294002 and then assayed for phosphorylated ERK, phosphorylated Akt, and survivin. As expected, U0126 (15 or 30 μmol/L) decreased phosphorylated ERK and LY294002 (20 or 40 μmol/L) decreased phosphorylated Akt (Fig. 3). Both inhibitors also induced a concentration-dependent decrease in survivin. Evidently, at the higher drug concentrations, survivin expression was completely abolished (Fig. 3). These results, along with earlier demonstrations that FTS treatment of a variety of glioblastoma multiforme cell lines inhibits the activation of ERK and Akt (9, 13), strongly suggested that the FTS-induced inhibition of survivin expression is associated with down-regulation of the Raf/MEK/ERK and PI3K/Akt/glycogen synthase kinase-3 Ras pathways.

As pointed out above, the expression of several regulators of survivin transcription, all of which are reportedly controlled by Ras signaling pathways (17–19), were altered by FTS in U87 cells (Fig. 1B). These regulators include E2F1 (17), p53 (18), and β-catenin (19). E2F1 binds to the survivin promoter and acts as a positive regulator of survivin expression (17). Studies by our group showed that FTS down-regulates E2F1 transcriptional activity (10). It was therefore reasonable to postulate that FTS down-regulates survivin by an E2F1-dependent mechanism. To investigate this possibility, we used U87 cells with forced E2F1 expression (U87-E2F1 cells), in which the high levels of E2F1 prevented FTS from inhibiting cell cycle progression (10). Unlike in U87 cells, in which the Ras-controlled E2F1 expression is disrupted by FTS (see also Fig. 4A, top; ref. 10), in U87-E2F1 cells, FTS had only a minor effect on the levels of E2F1 (Fig. 4A). We were therefore able to compare the effects of FTS on survivin levels in U87 cells and in U87-E2F1 cells, in which the amounts of E2F1 are not altered by FTS.

Two types of experiments were done. First, we determined the amounts of survivin by Western immunoblotting with anti-survivin antibody. As expected (17), the forced expression of E2F1 caused marked up-regulation of survivin (Fig. 4B). Treatment with 75 μmol/L FTS for 48 hours, however, caused a marked decrease in survivin not only in U87 cells but also in U87-E2F1 cells (Fig. 4B). Second, we cotransfected U87 cells with a 6-fold excess of E2F1 and GFP (to distinguish between transfected and nontransfected cells) and then did immunofluorescence analysis of controls and FTS-treated cells using anti-survivin antibody. The results of these experiments showed that there was more survivin (red fluorescence) in the nuclei of cells expressing excess E2F1 (cells labeled with green fluorescence) than in the nontransfected cells (without green fluorescence; Fig. 4C). The percentage of E2F1-expressing cells with relatively large amount of nuclear survivin was 89% and that of the nontransfected cells was 55% (Fig. 4C). Here too, treatment with FTS strongly decreased survivin both in cells overexpressing E2F1 and in the nontransfected cells (Fig. 4C). Taken together, these experiments showed that FTS induces down-regulation of survivin even under conditions, in which it does not cause a significant reduction in E2F1. It therefore seems that the effect of FTS on survivin expression might be mediated by its effect not only on E2F1 but also on other regulators of survivin.

Next, we examined the effects of FTS on two additional regulators of survivin: p53, which inhibits survivin gene expression (18), and β-catenin, which positively regulates survivin gene expression (19). We treated U87 cells with 75 μmol/L FTS for 24, 48, or 72 hours and then used Western immunoblotting, using anti-p53 and anti-β-catenin antibodies, to determine the amounts of p53 and β-catenin. The p53 protein showed a 2-fold increase 24 hours after FTS treatment and remained at this relatively high level 48 and 72 hours after treatment (Fig. 4D). This finding suggested that p53 could potentially be involved in the drug-induced down-regulation of survivin. The amount of β-catenin was not altered 24 hours after the treatment with FTS (Fig. 4D), a time point at which survivin levels were already markedly decreased (by 40%; Fig. 1D). However, a decrease in β-catenin was observed 48 hours after the FTS treatment (Fig. 4D). Thus, the early effect of FTS on survivin does not seem to be associated with β-catenin regulation.

Next, we examined whether the FTS-induced down-regulation of survivin is associated with the drug-induced death of U87 cells. For this purpose, we established U87 cell lines with forced expression of HA-tagged survivin and compared the effect of FTS on the death of U87 and of U87-survivin cells. Two U87-survivin cell lines were established, the one overexpressing HA-survivin (U87-survivin-H; Fig. 5A, top, left and right) and the other expressing HA-survivin at levels comparable with the levels of endogenous survivin (U87-survivin-L; Fig. 5A, bottom, left). Immunofluorescence analysis with anti-survivin antibody showed strong nuclear localization of survivin in the U87-survivin-H cells and no effect of FTS (75 μmol/L; 48 hours) on this localization or on the total amount of survivin in the cells (Fig. 5A, bottom). These results were confirmed by Western immunoblot analysis, which showed that the FTS treatment did not affect the amounts of the abundant HA-tagged survivin in these cells (Fig. 5A, top). As expected, the treatment caused a marked decrease in endogenous survivin in both U87 and U87-survivin-H cells (Fig. 5A, top). FTS treatment caused the death of U87 cells (>40% of the cells exhibited a subdiploid DNA content, sub-G₁ phase) but not of U87-survivin-H cells (7.5% dead cells; Fig. 5B, left). Consistent with these observations, live cells accounted for 40% of the FTS-treated U87 cells but 90% of the U87-survivin-H cells (Fig. 5B, right). Similar results were obtained with U87-survivin-L cells, which were partially protected from FTS-induced cell death (Fig. 5B, left and right). This showed that maintenance of HA-survivin at levels comparable with those of wild-type survivin is sufficient to attenuate the FTS-induced death of U87 cells. Taken together, these results then showed that the death of U87 cells induced by FTS is overridden by forced expression of survivin, suggesting that the down-regulation of survivin by FTS is associated with the drug-induced cell death.

To further substantiate this suggestion, we studied the effect of FTS on apoptosis in U87 cells because survivin, like other members of the inhibitor of apoptosis protein family of proteins, can act as an antiapoptotic factor, which, inter alia, inhibits caspase-3 and caspase-7 and was shown to inhibit caspase-9-dependent apoptosis (21–24). U87 cells were treated with 75 μmol/L FTS for 48 hours and then double stained with Hoechst to label the cell nuclei (blue) and with rabbit anti-survivin antibody followed by Cy3-conjugated anti-rabbit antibody (red, Cy3 fluorescence). Typical dual fluorescence images (Fig. 6A) show Hoechst staining of round intact nuclei of the control cells with large amounts of survivin. In marked contrast, Hoechst staining of the FTS-treated cells showed fragmented and condensed nuclei (and smaller amounts of survivin) typical of apoptotic nuclei (Fig. 6A). Quantitative analysis of the results showed that FTS induced an increase of 80% in apoptotic nuclei and a decrease of 85% in survivin (Fig. 6A). These experiments suggested that FTS induces apoptotic death of U87 cells.

To further support this possibility, we examined the effect of caspase inhibitors on the FTS-induced death of U87 cells. The results showed that zVAD (Z-VAD-FMK), a broad-spectrum inhibitor of caspases, or QVD (Q-VD-OPH), a pan-caspase inhibitor, exerted a marked protective effect against the FTS-induced death of U87 cells (Fig. 6B). Evidently, in the presence of these inhibitors, some 60% to 70% of cells survived. Treatment of the cells with a specific caspase-3 inhibitor (Z-DEVD-FMK) also resulted in their protection from FTS-induced cell death but to a smaller extent than that observed with the general caspase inhibitors (Fig. 6B). Taken together, these results suggested that FTS induces the caspase-dependent apoptotic death of U87 cells. Consistent with this possibility, we found that FTS induces cleavage of procaspase-3 and procaspase-7 and of their upstream activator procaspase-9, resulting in the appearance of the corresponding cleavage products, which are known as active caspases (Fig. 6C). In addition, we found that FTS induces cleavage of poly(ADP-ribose) polymerase, one of the death substrates that is cleaved by caspase-3 (Fig. 6C). Importantly, the FTS-induced cleavage of the caspases and of poly(ADP-ribose) polymerase was completely blocked by forced expression of survivin (Fig. 6C). Because release of cytochrome c from the mitochondria is essential for apoptosome formation, which leads to the activation of caspase-9, we next examined the effect of FTS on the proapoptotic molecules Bad and Bax, known to promote cytochrome c release. Bad is an “enabler” of cytochrome c release because it neutralizes Bcl2 or Bcl-xL, and the neutralization then enables Bax to activate cytochrome c release from the mitochondria (25). Bad is active in its nonphosphorylated form and is inactivated by Akt-mediated phosphorylation on Ser¹³⁶ (26). Consistent with the finding that FTS induces inhibition of PI3K activation and phosphorylation of Akt (9), we observed a decrease in phosphorylated Bad (Ser¹³⁶) in U87 cells that were incubated with 75 μmol/L FTS for 48 and 72 hours (Fig. 6C). We then did immunofluorescence experiments in the FTS-treated U87 cells using an anti-Bax NH₂-terminal antibody that recognizes the NH₂ terminus of the open, active conformer of Bax, leading to release of cytochrome c from the mitochondria (27). The results of these experiments showed positive staining with the anti-Bax NH₂-terminal antibody, indicating that FTS induces activation of Bax (Fig. 6D). Consistent with the observed activation of Bax and Bad, we found that FTS induces cytochrome c release from the mitochondria of U87 cells (Fig. 6D).

The results of this study showed that the Ras inhibitor FTS causes a decrease in the level of survivin transcripts in U87 glioblastoma multiforme cells (Fig. 1B and C), with a concomitant marked decrease in survivin protein (Fig. 1D). Decrease in survivin gene transcripts, together with the short half-life of survivin (which is rapidly degraded by the proteasome; ref. 28), can explain the observed reduction in survivin in FTS-treated U87 cells. Consistent with the notion that the decrease in survivin in FTS-treated U87 cells is mainly, if not entirely, attributable to decrease in the levels of survivin mRNA are the results obtained with U87-survivin cells, in which survivin expression is not under the control of transcription factors that bind to the survivin promoter. In these cells, FTS induced down-regulation of the endogenous but not of the forcibly expressed survivin (Fig. 5A). The effect of FTS on survivin was mimicked by dominant-negative Ras (Fig. 2C) and by inhibitors of the Raf/MEK/ERK and the PI3K/Akt pathways (Fig. 3).

These results are consistent with the conclusion that the observed FTS-induced inhibition of Ras and the consequent inhibition of activation of the Ras-dependent ERK and Akt (Fig. 1; ref. 9) are associated with the drug-induced down-regulation of survivin expression in U87 cells. Other studies showed, in addition, that MEK and PI3K inhibitors induce down-regulation of survivin in other cell lines (20). In line with this assertion, earlier studies showed that the major factors regulating survivin transcription positively (i.e., E2F1 and β-catenin; refs. 17, 19) or negatively (i.e., p53; ref. 18) are controlled by these Ras pathways. Moreover, in agreement with previous results in other tumor cell lines (29) and in glioblastoma multiforme (10), we showed that FTS induces down-regulation of E2F1 and up-regulation of p53 (Fig. 4A and D). In addition, FTS induced down-regulation of β-catenin (Fig. 4D). It seems, however, that neither of these regulators by itself is essential for down-regulation of survivin by FTS. This is because (a) FTS down-regulated survivin in glioblastoma multiforme cells with or without p53 mutation (Fig. 2A), (b) FTS down-regulated survivin expression in U87 cells with forced E2F1 expression (Fig. 4B), and (c) FTS down-regulation of survivin preceded the decrease in β-catenin (Fig. 4D). Thus, apparently, a Ras-regulated factor, as yet unknown, controls survivin transcription in glioblastoma multiforme.

Survivin has been implicated in the inhibition of apoptosis and is associated with reduced apoptotic index in human tumor cells (30). Survivin was found to be up-regulated in various human cancers, including neuroblastomas (31), bladder cancer (32), colon/colorectal cancer (33), gastric cancer (34), hepatocellular carcinomas (35), and human gliomas (36). Survivin, like other mammalian inhibitor of apoptosis proteins (37), can bind to caspase-3 and caspase-7 (21). Studies with a transgenic mouse model that selectively expresses survivin in the skin confirmed that this protein inhibits the intrinsic, caspase-9-dependent apoptotic pathway (22). Accumulating data point to the participation of survivin both in cell cycle progression during mitosis through its association with the mitotic spindle (38) and in blocking of apoptosis through its inhibitory effects on caspase-3, caspase-7, and caspase-9 (23, 24). Our results showed that the down-regulation of survivin by FTS in U87 cells is accompanied by caspase-mediated apoptotic cell death (Figs. 5 and 6), which is rescued by forced expression of survivin (Fig. 5). These results strongly suggested that the FTS-induced apoptosis in U87 cells is mediated, at least in part, by deregulation of survivin and facilitation of the intrinsic caspase-dependent pathways. Caspase inhibitors indeed attenuated the FTS-induced death of U87 cells (Fig. 6B). Consistent with this notion, previous studies showed that dominant-negative (T34A) survivin (39) or survivin antisense oligonucleotides (40) induce caspase-dependent apoptosis in glioblastoma multiforme. Down-regulation of survivin in other human tumor cell lines also resulted in increased apoptosis (41). From all of the above, it seems that inhibition of Ras can override deregulated survivin functions in cancer cells, thereby relieving the survivin-associated blockage (“braking”) of cell death.

The relief of survivin inhibition of intrinsic apoptotic pathways under certain circumstances or a genetic background seems to be sufficient to facilitate tumor cell death (42). It is most likely, however, that this effect operates because it is accompanied by mechanisms promoting activation of death pathways or preventing survival signals. Such mechanisms can operate both extracellularly and intracellularly. It is important to note that, because active Ras controls multiple targets, including intrinsic apoptotic pathways, the Ras inhibition in glioblastoma multiforme that leads to cell death (Figs 5 and 6) results both in inhibition of survivin expression and in activation of intrinsic apoptotic pathways. Inhibition of Ras and its pathways in U87 cells (9) was indeed accompanied by activation of Bad and Bax, leading to release of cytochrome c from the mitochondria (Fig. 6). Interestingly, microarray analysis enabled us to detect an increase (2.6-fold) in Bax in FTS-treated U87 cells.³

It should be emphasized that inhibition of Ras by FTS or by dominant-negative Ras leads to proteasomal degradation of HIF-1α (9) and at the same time to inhibition of survivin transcription. The loss of HIF-1α has numerous consequences, including glycolysis shutdown and promotion of cell death. Indeed, ATP production is dramatically reduced in FTS-treated U87 cells (9). Recent experiments showed that inhibition of HIF-1α by phosphorothioate antisense HIF-1α oligonucleotide induces p53-independent apoptosis in U87 cells (12). Other studies showed that knockout of HIF-1α gene expression via antisense HIF-1α or by siRNA treatment induces apoptosis of human squamous cell carcinoma (43). These effects might be associated with the lack of glycolysis in tumors cells, in which HIF-1α was down-regulated; recent experiments showed that direct inhibition of glycolysis in human lymphoma and leukemia cells caused dephosphorylation of Bad, leading to translocation of Bax to the mitochondria, loss of mitochondrial membrane integrity, release of cytochrome c, and activation of apoptotic cascades (44). Such a mechanism whereby mitochondrial membrane integrity is lost may also operate in FTS-treated U87 cells, in which FTS down-regulates HIF-1α and blocks glycolysis. Consistent with this possibility, we showed that FTS inhibits PI3K and Akt phosphorylation, inhibits Bad phosphorylation, and induces Bax translocation to the mitochondria (Fig. 6). In addition, we showed earlier that FTS down-regulates expression the HIF-1α-controlled glucose transporter Glut1 (9). Glut1 was shown to prevent metabolism-sensitive conformational changes in Bax and thus to inhibit cell death (45). Other studies showed that expression of HIF-1α closely correlates with expression of survivin (46, 47). Thus, reduced expression of HIF-1α seems to induce Bax activation and at the same time to reduce expression of survivin. More experiments are needed to tell whether the FTS-induced down-regulation of HIF-1α (9) and of survivin are dependent events.

In conclusion, we propose that the strong apoptotic cell death induced by FTS in glioblastoma multiforme could be the result of both relief of the apoptosis “brake” (survivin) and induction of intrinsic apoptotic pathways possibly associated with the disappearance of HIF-1α (9). The results described here nonetheless provide a strong rationale for the potential use of FTS as treatment modality for glioblastoma multiforme.

We thank the Arison family's donation to the Center of DNA Chips in Pediatric Oncology, Chaim Sheba Medical Center and S.R. Smith for editorial assistance.