|Home | About | Journals | Submit | Contact Us | Français|
Oncogenic potential is associated with translational regulation, and the prevailing view is that oncogenes use mTOR-dependent pathways to up-regulate the synthesis of proteins critical for transformation. In this study, we show that RalA, a key mediator of Ras transformation, is also linked to the translational machinery. At least part of this linkage, however, is independent of mTOR and acts through RalBP1 to suppress cdc42-mediated activation of S6 kinase and the translation of the antiapoptotic protein FLIPS. This action, rather than contributing to transformation, opens a latent tumor-suppressive mechanism that can be activated by tumor necrosis factor-related apoptosis-inducing ligand. These results show that the translational machinery is linked to tumor suppression as well as cell-proliferative pathways and that the reestablishment of cell death pathways by activation of the Ral oncogenic program provides a means for selective therapeutic targeting of Ral-driven malignancies.
Controlled proliferation depends on accurate and timely regulation of cellular protein levels and activity. To accommodate this need, cells have evolved a variety of mechanisms to control gene expression, the most proximal of which regulate translation. The process of translation is highly regulated, beginning at the rate-limiting step of initiation (36). For capped mRNAs, the binding of eukaryotic initiating factor 4E (eIF-4E) to the mRNA cap structure facilitates the formation of the eIF-4F complex, the melting of secondary structures near the 5′ cap, and the ribosomal recruitment/loading of the mRNA onto polyribosome complexes (polysomes) (13, 14). For uncapped mRNAs with 5′ terminal oligopyrimidine sequences, the initiation process may be facilitated by S6K1-mediated phosphorylation of S6, a component of the 40S ribosome (10, 12-14), although recent evidence suggests that translational regulation of 5′ terminal oligopyrimidine mRNAs occurs even in the absence of S6K1 (43). The activities of both eIF-4E and S6K are closely regulated by a variety of factors, the most intensely studied being mTOR. mTOR phosphorylates S6K1, allowing the recruitment of PDK-1, which in turn activates S6K1 by Thr-229 phosphorylation (14). mTOR also phosphorylates the three isoforms of 4E-BP, inhibiting the ability of these proteins to compete with eIF-4G for eIF-4E binding and block translation initiation (11). Given the importance of mTOR in regulating translation, it is not surprising that mTOR and the translational process itself are co-opted by transforming oncogenes. Oncogenic activation of the Ras pathway, for example, stimulates translation in at least two ways: phosphatidylinositol 3-kinase (PI3K), via the extensively described Akt-mTOR pathway, stimulates translation by activating eIF-4E and S6K (14), while in some circumstances, the Raf-extracellular signal-regulated kinase pathway stimulates translation via MNK-mediated phosphorylation of eIF-4E (2, 34, 45). The ability of multiple oncogenic pathways to stimulate translation as well as the observation that the mTOR pathway is hyperactive in many tumors, has led to the suggestions that enhanced protein synthesis provides a growth advantage to tumors and that regulators of translation, in particular mTOR, may be promising therapeutic targets (17).
In addition to stimulating translation, however, growth-promoting oncogenic pathways also paradoxically activate latent tumor suppressor pathways that leave tumors vulnerable to cell death. Raf activation, in addition to activating eIF-4E, also induces the expression of p16/ARF, indirectly activating Rb and/or p53, which in turn influences the expression/activity of controllers of senescence, including p21 and PML (24, 40). The activation of the Ras/Raf/extracellular signal-regulated kinase pathway has also been shown (perhaps by myc stabilization) to transcriptionally up-regulate the death receptors DR4 and DR5 and stimulate caspase-8 recruitment to the death-induced signaling complex (DISC), all of which sensitize cells to apoptosis induced by the death ligand TRAIL (29, 37, 46, 47). Enhanced translation mediated by oncogenic activation is therefore countered by an “oncogenic checkpoint” which, when activated, sensitizes cells to death signals in their environments and limits tumor growth.
While the Raf and PI3K pathways have been linked to both control of translation and activation of innate tumor-suppressive mechanisms, relatively little is known about a third recently identified effector of Ras oncogenesis, Ral. The Ral proteins (RalA and RalB) are small GTPases whose activity is controlled by the RalGEF family of proteins, of which at least three members (RalGDS, Rgl, and Rlf) are known (48). Activated Ras binds RalGDS via a Ras-binding domain and recruits it to the membrane, where it stimulates an exchange of GDP for GTP on both forms of Ral (22, 28, 35). Activated ral molecules in turn interact with a myriad of Ral effectors, including RalBP1 (a cdc42 GTPase-activating protein) (4, 21), subunits of the exocyst complex (Sec5 and Exo84) that direct vesicles to the plasma membrane (28), the transcription factor Zonab (38), and the actin-binding protein filamin (30). While RalA/RalB activation plays a role in membrane trafficking, actin organization, and gene expression, the expression of a Ras point effector mutant selective for Ral GDS activation (H-Ras12V E37G) also transforms immortalized human fibroblasts, embryonic kidney epithelial cells, and astrocytes (18). Consistent with this idea, RalGDS was shown to be required for tumor formation in a mouse model of skin carcinogenesis (16); this requirement is perhaps related to Ral's ability to stimulate the Jun N-terminal protein kinase/stress-activated protein kinase pathway and suppress apoptosis and enhance cell survival or alternatively related to interactions between Ral and cytoskeletal reorganizing proteins, such as RalBP1, Rac1, and cdc42. The relative contributions of RalA and RalB to cellular transformation remain uncertain, although the proteins clearly differ in function (for example, RalA binds more effectively to the exocyst components than RalB does) (41). The suppression of RalA blocks Ras- and Ral-mediated transformation of cultured fibroblasts (23), while the suppression of RalB leads to apoptosis (6), suggesting that both RalA and RalB may play key roles. While these studies firmly establish the Ral pathway as a key contributor to Ras-mediated tumorigenesis in mouse and humans, they leave open the question of whether the Ral pathway, like other oncogenic pathways, is linked to the translational machinery and innate tumor-suppressive mechanisms. We address these possibilities in the present study and show that the RalA pathway is linked to both the translational control and the tumor-suppressive pathways. Unlike other oncogenes, however, the ability of RalA to regulate translation is independent of its transforming ability but directly responsible for its ability to sensitize glioma cells to the activation of the extrinsic death pathway by TRAIL. These results define an mTOR-independent link between RalA and translation and identify a means by which Ral-regulated malignancies may be attacked therapeutically.
Immortalized or Ras-transformed human astrocytes were generated and cultured as described previously (42). Human xenografted glioblastoma multiforme (GBM) cells were obtained from the UCSF Brain Tumor Research Center Tissue Bank and cultured as previously described (31). Human recombinant TRAIL and rapamycin were purchased from Sigma and Cell Signaling Technology, respectively, and dissolved in dimethyl sulfoxide.
Cells (1 × 107) were resuspended in a 1:1 mixture of phosphate-buffered saline-Matrigel and injected subcutaneously into the flanks of rag1−/− immunodeficient mice (C57BL6/129Sv), after which tumor volumes were determined at regular intervals as described previously (23).
Cell lysates were prepared and carried through Western blot analysis as previously described (31) by using antibody against alpha tubulin, p70 S6K1, or phospho-S6K1 (Thr-389) (all mouse polyclonal antibodies; Santa Cruz Biotechnology), phopho-S6 (rabbit polyclonal antibody; Santa Cruz Biotechnology), FLIPL and FLIPS (goat monoclonal antibody; Santa Cruz Biotechnology), FADD (mouse monoclonal antibody; Abcam), RalBP1 (rabbit monoclonal antibody; Cell Signaling Technology), and cdc42, caspase 3, and caspase 8 (rabbit polyclonal antibodies; Cell Signaling Technology). Bound antibody was detected with mouse anti-goat immunoglobulin G (IgG), goat anti-rabbit IgG, or goat anti-mouse IgG (Santa Cruz Biotechnology) by using ECL Western blotting detection reagents (Amersham Pharmacia Biotech, Inc.). The expression of alpha tubulin was used to verify equal loading in all studies. The extent of apoptosis in cultures (attached and floating cells) was determined by fluorescence-activated cell sorter analysis (sub-G1 DNA content), with measurements verified by annexin V-propidium iodide staining as previously described (5).
Retroviral constructs encoding FLIPS or FLIPL were provided by L. Bin (3). An expression vector encoding hemagglutinin-tagged wild-type (WT) p70 S6K1 (pRK7/HA-S6K1) was kindly provided by Robert Abraham (Burnham Institute, San Diego, CA). Constructs encoding Ras point effector mutants (H-Ras12V E37G, T35S, or Y40C) or WT RalA were provided by Pablo Rodriguez-Viciana (UCSF Cancer Center, San Francisco, CA), while the WT cdc42 and constitutively active (Q61L) and kinase-dead (T17N) cdc42 constructs were provided by David Stokoe (UCSF Cancer Center San Francisco, CA). Constructs encoding WT RalBP1, RalBP1Δ65-80, or RalBP1Δ154-219 were provided by Sanjay Awasthi (University of Texas—Arlington) (51), while a construct encoding RalBP1 with an N-terminal deletion of the GAP domain (RalBP1ΔGAP) was provided by Pablo Rodriguez-Viciana (9). Pools of productively infected cells (obtained by selection with 1 μg/ml neomycin, 7 days, or 9 μg/ml puromycin, 7 days) (38) were used for further analysis. In cells expressing multiple constructs, all retroviral infections and selections were done serially. For short interfering RNA (siRNA) studies, 200 nM FLIPS-, FLIPL- (33) or RalBP1-targeted siRNA (Ambion; identification no. 18693) or 1 μM p70 S6 kinase SMARTpool siRNA or scramble siRNA (Dharmacon, Lafayette, CO) was transfected into cells and protein levels were analyzed 1 to 4 days later.
For the analysis of components of the assembled DISC, cells were incubated with TRAIL (800 ng/ml, 24 h) and lysed, after which the DISC was immunoprecipitated using an anti-FADD antibody and levels of components were assessed by Western blot analysis. Immunoprecipitations carried out using a nonspecific normal mouse IgG antibody were included as negative controls, as were analyses of cells to which TRAIL (800 ng/ml) was added following lysis.
Cells were lysed, preincubated with vehicle, GDP (negative control to inactivate cdc42-GTP), or a nonhydrolyzable form of GTP (GTPγS, positive control to activate cdc42-GTP), and incubated with an agarose-conjugated p21-binding domain (PBD) of PAK-1, a substrate of the activated (GTP bound) form of cdc42, or an agarose-conjugated nonspecific substrate (normal rabbit IgG). Substrate-bound (activated) cdc42 was then eluted, and levels were assessed by Western blot analysis by using a cdc42 antibody. Levels of total cdc42 were also determined in lysates prior to incubation with PAK-1 PBD.
The fractionation of cells by sucrose density gradient centrifugation was performed as previously described (20). The gradient was divided into 48 fractions, each of which was analyzed for absorbance at 260 nm then pooled into a total of 12 fractions. RNA from each fraction or from total cell lysates was spiked with 0.5 μg exogenous Drosophila ribosomal protein L3 (RPL3) mRNA (Ambion) (to control for losses of mRNA during purification) before purification using TRIzol reagent (Invitrogen; QIAGEN). Northern blot analyses were carried out using probes generated by reverse transcription-PCR.
Previous studies showed that the expression of a Ras effector mutant selective for the Ral pathway (H-Ras12V E37G) or of activated RalA itself was sufficient to transform immortalized human fibroblasts, embryonic kidney epithelial cells, and astrocytes (18, 23). Consistent with these results, the expression of the Ral-selective G37 mutant (but not of T35S or Y40C mutants selective for Raf or PI3K activation, respectively) was sufficient in our studies to allow immortalized human astrocytes (E6/E7/hTERT) to grow in soft agar (not shown) and form tumors following injection into immunodeficient animals (Fig. (Fig.1A).1A). Similar albeit smaller effects were noted in immortalized astrocytes retrovirally infected with a construct encoding WT RalA itself (Fig. (Fig.1A).1A). Because Ras-mediated transformation sensitizes fibroblasts to the extrinsic cell death activator TRAIL (29), we also examined TRAIL sensitivity in the various Ras-pathway-modulated astrocytes. As shown in Fig. Fig.1B,1B, the expression of only constructs that led to cellular transformation (H-Ras12V, G37, and WT RalA) sensitized immortalized astrocytes to TRAIL-induced apoptosis. This sensitization was not the result of changes in DR4/DR5 receptor expression as is noted following myc transformation of fibroblasts (see Fig. S1 in the supplemental material) (39, 46), nor was it associated with TRAIL-induced formation of a FADD antibody immunoprecipitable DISC complex, which was noted in both TRAIL-resistant E6/E7/hTERT cells and TRAIL-sensitive Ras or G37 cells (Fig. (Fig.1C).1C). Rather, TRAIL sensitization was associated with the conversion of caspase-3 and caspase-8 from uncleaved forms found in whole-cell lysates (Fig. (Fig.1C,1C, lanes 3) to cleaved/activated forms in the DISC (lanes 4 versus lanes 3) exclusively in TRAIL-sensitive cells.
FLIP, a protein that exists in long (55-kDa, FLIPL) and short (28-kDa, FLIPS) forms, is a modulator of DISC-associated caspase cleavage and TRAIL sensitivity (19), and the suppression of FLIPS levels sensitizes cells to TRAIL-induced apoptosis (31). To determine whether FLIP expression was associated with RalA activation or played a role in Ral-mediated enhancement of TRAIL sensitivity, we first compared the expression of activated RalA and FLIP in Ras- or Ral-transformed TRAIL-sensitive cells to that in TRAIL-resistant T35S or Y40C mutant-expressing cells. While levels of total RalA were essentially constant across all cell lines (except the E6/E7/hTERT cells manipulated to overexpress Ral A), RalA-GTP levels were significantly higher in Ras- or Ral-transformed TRAIL-sensitive cells (E6/E7/hTERT/Ras, G37, or RalA) than those in the TRAIL-resistant cell lines (E6/E7/hTERT, C40, and S35) (Fig. (Fig.1D).1D). Consistent with the idea that RalA-mediated increases in TRAIL sensitivity might be associated with altered FLIP expression, FLIPS levels (but not FLIPL levels) were significantly lower in the TRAIL-sensitive cells (E6/E7/hTERT/Ras, G37, or RalA) with high RalA activation than those in the TRAIL-resistant cell lines (E6/E7/hTERT, C40, and S35) with low RalA activation (Fig. (Fig.1E).1E). To further assess the importance of both forms of FLIP in controlling TRAIL resistance, TRAIL-sensitive Ras- or Ral-transformed cells were stably infected with a blank retrovirus or with retroviral constructs encoding FLIPS or FLIPL and assessed for levels of FLIPS/FLIPL protein and TRAIL sensitivity. Cells expressing the FLIPL-encoding construct exhibited a twofold increase in FLIPL expression but underwent TRAIL-induced apoptosis to the same extent as did control cells that received a blank construct (Fig. (Fig.1F,1F, bars 5 versus bars 2 and 3). Cells expressing the FLIPS-encoding construct exhibited twofold increases in FLIPS levels but exhibited significantly less TRAIL-induced apoptosis than did cells that received an empty construct (Fig. (Fig.1F,1F, bars 4 versus bars 3). In converse experiments, TRAIL-resistant cells expressing the Y40C Ras mutant selective for PI3K (C40 cells) were transfected with siRNA targeting FLIPS or FLIPL, after which TRAIL-induced apoptosis was assessed. While both siRNAs selectively decreased the expressions of their targets (Fig. (Fig.1G),1G), only FLIPS-targeted siRNA significantly increased TRAIL-induced apoptosis (Fig. (Fig.1H).1H). These results show that the activation of the RalA pathway suppresses levels of FLIPS, that the suppression of FLIPS levels sensitizes cells to TRAIL-mediated apoptosis, and that FLIPS links the Ral oncogenic pathway to the tumor-suppressive mechanism activated by TRAIL.
The means by which RalA is linked to the modulation of FLIPS levels have not been examined, although a variety of downstream targets of Ral have been suggested. To define the pathway coupling RalA to apoptotic sensitization, we first examined the role of the Ral target RalBP1. TRAIL-sensitive Ras- or Ral-transformed cells were incubated with siRNA targeting RalBP1, after which the effects of RalBP1 suppression on FLIPS expression and TRAIL sensitivity were monitored. E6/E7/hTERT/Ras, G37, and RalA-transformed cells transfected with siRNA targeting RalBP1 exhibited prolonged suppression of RalBP1 protein levels (Fig. (Fig.2A),2A), increased levels of FLIPS (Fig. (Fig.2A),2A), and decreased TRAIL sensitivity relative to cells transfected with a control siRNA (Fig. (Fig.2B,2B, top panel). In converse experiments, TRAIL-resistant nontransformed E6/E7/hTERT, C35, and C40 astrocytic cell lines exhibiting high levels of FLIPS were retrovirally infected with constructs encoding WT RalBP1 or mutant forms of RalBP1 incapable of binding ATP and serving as a membrane transporter (Δaa65-80), incapable of incorporating into the membrane (Δaa159-214) (46), or lacking GAP activity and incapable of acting upon CDC42 (ΔGAP) (9), after which the effects on FLIPS expression and TRAIL sensitivity were monitored. The overexpression of WT RalBP1 or of a mutant form incapable of binding ATP significantly suppressed levels of activated cdc42 (but not of total cdc42) (Fig. (Fig.2C),2C), suppressed levels of FLIPS, and enhanced TRAIL sensitivity (Fig. (Fig.2D,2D, lanes/bars 4 and 6 versus lanes/bars 3), while the overexpression of the RalBP1 mutants incapable of incorporating into the membrane (Δaa159-214) or lacking GAP activity (ΔGAP) did not alter levels of active cdc42, FLIPS, or TRAIL sensitivity relative to vector controls (Fig. 2C and D, lanes/bars 5 versus lanes/bars 3). These results show that RalBP1 serves in a membrane-bound, GAP-dependent manner to link RalA activation to the suppression of cdc42 activation, the suppression of FLIPS expression, and enhanced sensitivity to TRAIL-induced apoptosis.
To further examine the potential link between RalBP1, cdc42, and FLIPS expression, we first compared levels of activated cdc42 (cdc42-GTP) among Ras-modulated cell lines. As in Fig. Fig.2C,2C, cell lysates were preincubated with vehicle, GDP (a negative control to deplete cdc42-GTP), or a nonhydrolyzable form of GTP (positive control to enhance cdc42-GTP levels), followed by incubation with an agarose-conjugated substrate of the activated (GTP bound) form of cdc42 (PAK-1 PBD). Levels of eluted cdc42-GTP and total cdc42 were then assessed by Western blot analysis. TRAIL-sensitive Ras- or Ral-transformed cells with low levels of FLIPS (Fig. (Fig.1E)1E) and low levels of activated RalA (Fig. (Fig.1D)1D) also had low levels of active cdc42 (Fig. (Fig.3A,3A, lanes 5, 6, and 9), while TRAIL-resistant E6/E7/hTERT, S35, and C40 cells with high levels of FLIPS and activated RalA had higher levels of cdc42-GTP (lanes 4, 7, and 8, respectively). The Ras- or Ral-transformed cells transfected with siRNA targeting RalBP1 that exhibited significantly increased levels of FLIPS and increased TRAIL resistance (Fig. (Fig.2B)2B) also exhibited increased levels of activated cdc42 (Fig. (Fig.2B,2B, bottom panel, lanes/bars 5 versus lanes/bars 4), while the E6/E7/hTERT, C35, and C40 cells retrovirally infected with WT or mutant RalBP1 constructs that exhibited suppressed levels of FLIPS, decreased levels of activated cdc42, and increased TRAIL sensitivity (Fig. 2C and D, lanes 4 and 6) also exhibited suppressed levels of activated cdc42 (Fig. (Fig.2C2C).
To more directly determine whether cdc42 activity modulated FLIPS levels and TRAIL response, TRAIL-sensitive Ras- or Ral-transformed cells with high levels of active RalA and low levels of cdc42-GTP and FLIPS (E6/E7/hTERT/Ras, G37, and E6/E7/hTERT/RalA) were retrovirally infected with a blank retroviral construct (pLXSN-neo) or a construct encoding either constitutively active or dominant-negative cdc42 (Q61L and T17N, respectively) (1), after which stably expressing cells were selected and effects on FLIPS levels and TRAIL sensitivity were assessed. Ras- or Ral-transformed cells infected with the active cdc42 construct had increased levels of FLIPS (but not FLIPL) relative to those of controls (Fig. (Fig.3B,3B, lanes 4 versus lanes 2), and the DISC complex immunoprecipitated from these cells following exposure to TRAIL had increased levels of FLIPS and decreased levels of cleaved caspase-8/caspase-3 (Fig. (Fig.3C,3C, compare lanes 6 to lanes 4 and 5). The increased FLIPS expression and decreased DISC caspase cleavage was also accompanied by changes in TRAIL sensitivity, with both the Ras- and Ral-transformed cells exhibiting decreased TRAIL-induced apoptosis relative to that of control cells or cells expressing the dominant-negative (T17N) cdc42 construct (Fig. (Fig.3D,3D, bars 4 versus bars 3 or 5, respectively). These results show that cdc42 activity is suppressed by Ras/Ral/RalBP1 activation and that decreases in cdc42 activity lead to decreased levels of FLIPS and enhanced TRAIL sensitivity.
Because FLIPS protein levels can be controlled by the distribution of the FLIPS mRNA between translating polyribosomes and nontranslating monosomes (31), we used sucrose gradient density centrifugation in combination with Northern blot analysis to assess the ability of cdc42 to control ribosomal distribution of FLIPS mRNA and Ral-enhanced TRAIL sensitivity. Lysates from TRAIL-resistant E6/E7/hTERT cells, TRAIL-sensitive Ral-transformed cells (G37 or Ral), or the same cells made TRAIL-resistant by the introduction of a construct encoding constitutively active cdc42 were layered on a continuous 5 to 70% sucrose gradient and centrifuged. Fractions were then collected and analyzed for ribosomal distribution (see Fig. S2 in the supplemental material), spiked with Drosophila RPL3 RNA (as an internal control), and analyzed for FLIPS, FLIPL, and RPL3 RNA levels. While FLIPL mRNA associated with both polysomal and monosomal fractions in a pattern that was independent of TRAIL sensitivity (not shown), FLIPS mRNA was selectively associated with the nontranslating monosomes in TRAIL-sensitive G37- or Ral-transformed cells but with the translating polysomes in TRAIL-resistant parental E6/E7/hTERT cells (Fig. (Fig.3E).3E). Furthermore, the overexpression of constitutively active cdc42 in the G37- or Ral-transformed cells shifted the distribution of FLIPS mRNA from the monosomal fraction to the polysomal fraction relative to control cells (Fig. (Fig.3E),3E), consistent with the increased levels of FLIPS protein noted in these cells. These results suggest that Ral/RalBP1 activation suppresses cdc42 activity, which in turn controls FLIPS mRNA distribution and translation, and by doing so controls FLIPS protein levels and ultimately TRAIL sensitivity.
Because mTOR is a target of Ras and can control ribosomal distribution of FLIPS mRNA (31), we examined the possibility that mTOR was involved in the Ras/Ral/RalBP1/cdc42 pathway that suppresses FLIPS mRNA translation and enhances extrinsic apoptosis. For these studies, E6/E7/hTERT/Ras, G37, and Ral cells overexpressing constitutively active cdc42 (and subsequently exhibiting high levels of FLIPS mRNA in translating polysomal fractions, high levels of FLIPS, lack of TRAIL-induced caspase cleavage in the DISC, and high TRAIL resistance) were incubated with the mTOR inhibitor rapamycin, after which the effects on Ral-related signal transduction were assessed. Although rapamycin exposure had no effect on cdc42-driven FLIPS protein overexpression (Fig. (Fig.3B,3B, lanes 5 versus lanes 4), it altered events in the DISC of Ras-or Ral-transformed cells, decreasing levels of FLIPS in the DISC and increasing cleavage of caspase-8/caspase-3 (Fig. (Fig.3C,3C, lanes 7 versus lanes 6), while reversing cdc42-mediated TRAIL sensitivity (Fig. (Fig.3F,3F, compare bars 7 to bars 5 and bars 3). These results show that while mTOR does not alter the ability of the Ral/RalBP1/cdc42 pathway to control FLIPS levels, it may modulate the ability of the FLIPS protein, once created, to localize to the DISC complex and to regulate the apoptotic process.
Because cdc42 activation increases FLIPS translation and confers TRAIL resistance, we considered the possible involvement of S6K1, a translational regulator and known downstream target of cdc42. Initial Western blot analysis showed that TRAIL-sensitive cells (E6/E7/hTERT/Ras, G37, and E6/E7/hTERT/Ral) with high levels of activated RalA and low levels of activated cdc42 and FLIPS also had lower levels of pS6K and pS6 than did TRAIL-resistant cells (E6/E7/hTERT, S35, and C40) (Fig. (Fig.4A).4A). The introduction of an siRNA targeting RalBP1 also suppressed RalBP1 levels (relative to sham-transfected cells or cells transfected with a scrambled siRNA) while simultaneously increasing levels of pS6K and pS6 in three different TRAIL-sensitive cell lines with high levels of RalA activation (Fig. (Fig.4B).4B). Furthermore, the ability of siRNA targeting RalBP1 to enhance pS6K and pS6 levels was independent of mTOR as the incubation of the above-described cells with rapamycin had no effect on the ability of the RalBP1 siRNA to enhance pS6K and pS6 levels (Fig. (Fig.4B).4B). We therefore performed further studies in which Ral-transformed cells were retrovirally infected with a construct encoding either constitutively active cdc42 or wild-type p70 S6K1. Levels of S6K1, phosphorylated (active) S6K1, phosphorylated S6, cdc42, active cdc42-GTP, FLIPS, and FLIPL were then assessed in untreated or rapamycin-treated cells, as was the ribosomal distribution of the FLIPS mRNA. The same cells were also incubated with either TRAIL or rapamycin followed by TRAIL, after which the extent of apoptosis was assessed. Cells that received the cdc42 construct (Fig. (Fig.4C,4C, bottom panel, lane 8) exhibited increased levels of cdc42, cdc42-GTP, pS6K, pS6, and FLIPS relative to those of the blank vector E6/E7/hTERT/Ras cells (lane 1/2) and also exhibited a shift in FLIPS mRNA from the monosomal to the polysomal fraction (Fig. 4C and D). These cells were also more TRAIL resistant than were vector control cells (Fig. (Fig.4C,4C, top panel, bar 8 versus bars 1 and 2). This degree of TRAIL resistance could be reversed by pretreatment with rapamycin (Fig. (Fig.4C,4C, top panel, bar 9 versus bar 8), which did not change FLIPS levels (Fig. (Fig.4C,4C, bottom panel, lanes 8 and 9) or polysomal FLIPS mRNA distribution (Fig. 4C and D, groups 8 and 9), but rather likely influenced the localization of FLIPS to the DISC. Cells that received the S6K1 construct did not have increased levels of cdc42 or cdc42GTP, but did exhibit increased levels of S6K1, pS6K, pS6, and FLIPS relative to blank vector cells (Fig. (Fig.4C,4C, lane 5 versus lane 4) and also exhibited a shift in FLIPS mRNA to the polysomal fraction (Fig. 4C and D, group 5 versus group 4). These cells, like those expressing the cdc42 construct, were also more TRAIL resistant than were vector control cells (Fig. (Fig.4C,4C, top panel, bar 5 versus bar 4), and this degree of TRAIL resistance could also be reversed by pretreatment with rapamycin independently of FLIPS levels and FLIPS mRNA distribution (Fig. 4C and D).
To further verify that S6K was a key link between cdc42 and FLIPS, E6/E7/hTERT/Ras cells retrovirally infected with a construct encoding cdc42 were also transfected with scramble siRNA or siRNA targeting S6K1, after which parameters associated with TRAIL-induced apoptosis were measured. The introduction of S6K1-targeted siRNA had no effect on cdc42 or cdc42GTP levels but significantly reduced S6K, pS6K, pS6, and FLIPS levels (Fig. (Fig.4C,4C, top panel, lane 11 versus lane 10), shifted FLIPS mRNA to the monosomal fraction (Fig. (Fig.4D,4D, groups 10 and 11), and sensitized cells to TRAIL-induced apoptosis as effectively as did the inhibition of the cdc42 pathway by Ral (Fig. (Fig.4C,4C, bars 10 and 11). These results show that cdc42-mediated activation of S6K1 causes an mTOR-independent redistribution of FLIPS mRNA to polyribosomes, increasing FLIPS protein levels and TRAIL resistance. Ral pathway activation, by virtue of its ability to suppress cdc42 activity, translationally suppresses FLIPS levels and sensitizes cells to TRAIL-induced apoptosis.
Finally, to determine whether the relationship between cdc42, S6K1, FLIPS, and TRAIL sensitivity was applicable to primary human GBM in addition to Ras/Ral-transformed astrocytes, primary human GBM passaged as in vivo xenograft and previously shown to be TRAIL sensitive by virtue of low levels of FLIPS were stably infected with a blank vector or a vector encoding either constitutively activated (Q61L) or mutant dominant-negative (T17N) cdc42, after which levels of cdc42, phosphorylated (active) S6K1, total S6K, FLIPS, and alpha tubulin (as a loading control) were assessed by Western blot analysis in untreated or rapamycin-treated cells. The same cells were also incubated with either TRAIL or rapamycin followed by TRAIL, after which the extent of apoptosis was assessed. Cells that received the constitutively active cdc42 construct (Fig. (Fig.5,5, bottom panel, lane 7) exhibited increased levels of cdc42, pS6K, and FLIPS relative to blank vector cells (lane 5) or cells that received dominant-negative cdc42 (lane 6) and were also more TRAIL resistant (Fig. (Fig.5,5, top panel, bars 7 versus bars 5 and 6). As in Ras/RalA-transformed astrocytes, this degree of TRAIL resistance could be reversed by pretreatment with rapamycin (Fig. (Fig.5,5, top panel, bars 8 versus bars 7), which did not change FLIPS levels (Fig. (Fig.5,5, bottom panel, lanes 7 and 8), but rather likely influenced the localization of FLIPS to the DISC. These results as a whole suggest that the ability of cdc42 to suppress S6K activation, to translationally suppress FLIPS levels, and to sensitize cells to TRAIL-induced apoptosis is not only unique to manipulated cells in culture but also applies to primary brain tumors as well.
Oncogenic lesions that enhance translation also frequently activate innate tumor suppressor pathways (25). Because the Ral pathway negatively controls translation, we questioned whether this ability was critical for Ral-induced transformation. To address this point, we determined whether the genetic manipulations that altered control of the extrinsic apoptotic pathway in Ral-transformed cells also altered Ral-mediated transformation. While the overexpression of cdc42 in Ras- or Ral-transformed (G37 and RalA) cells shifted FLIPS mRNA ribosomal distribution, translationally up-regulated FLIPS levels, and increased TRAIL resistance, cdc42 expression had no effect on the ability of the transformed cells to grow in soft agar (not shown) or in immunodeficient hosts (Fig. (Fig.1A).1A). These results suggest that the Ral pathway is linked to the control of translation but that this linkage neither is critical for nor interferes with Ral-mediated transformation.
Although oncogenic pathways frequently co-opt the translational machinery for their own benefit, they also paradoxically open tumor-suppressive mechanisms that can lead to their own demises. The present study shows that the RalA oncogenic pathway fits this mold, with a few novel twists. Like other oncogenic pathways, RalA activation hijacks the translational machinery. This translational co-option, however, is not the driving force behind the transformation process, but rather, by suppressing the translation of the extrinsic apoptotic pathway inhibitor FLIPS, is itself the key that opens a pathway to cell death.
The present work links the RalA pathway to control of the extrinsic apoptotic pathway activated by TRAIL. The mitogen-activated protein kinase pathway, but not the Ral or PI3K pathways, has been previously linked to the stabilization of myc, increased DR5 expression, increased caspase activation, and enhanced TRAIL-induced apoptosis in fibroblasts and HEK cells (29, 37, 46, 47). In the present study, there was no indication of RalA-induced DR5 up-regulation or myc stabilization (unpublished data), although RalA but not a Raf-selective T35S Ras effector mutant did increase caspase activation. These differences may be a consequence of different TRAIL exposure conditions used or alternatively may reflect tissue-specific differences in the control of the extrinsic apoptotic pathway and subtle differences in the balance between the multiple factors that set the apoptotic threshold. The finding that Ras, Raf, Ral, and myc all suppress TRAIL-induced caspase activation in multiple cell lines under multiple conditions (29, 37, 39, 46, 47), however, suggests that the regulation of DISC function, including that by FLIP up-regulation, may be a common means by which oncogenic pathways regulate the extrinsic apoptotic pathway. While FLIPS regulation can be accomplished transcriptionally, as has been previously reported for myc (37), the ability of the Ral pathway to do so translationally represents a novel means of controlling the extrinsic cell death pathway.
The means by which the RalA pathway is linked to the translational machinery is novel, as is the manner in which this oncogenic pathway impacts translation. The present work shows that RalA is linked to the control of FLIPS translation not via mTOR, the most commonly identified and extensively studied regulator of translation, but instead via a second GTPase, cdc42, and by the ability of cdc42 to regulate the activity of the mTOR target S6K1. Our observation that cdc42 activates S6K1 is consistent with previous reports that suggest that cdc42 coimmunoprecipitates with and activates S6K1 (7, 8). How a GTPase such as cdc42 activates S6K1, however, remains unclear, although previous studies using isoprenylation-deficient cdc42 mutants suggest that membrane targeting of cdc42 is critical for S6K1 activation (7). Membrane localization of the cdc42-S6K complex may allow other membrane-associated proteins such as PDK1 to activate S6K1, although it is clear that at least part of the ability of cdc42 to activate S6K1 is distinct from that of PDK1 (26). The observation that both RalBP1 (this study) and cdc42 (32) function in a membrane-associated manner is consistent with the idea that RalBP1, cdc42, and S6K may be brought into intimate contact by RalA activation in the cell membrane, after which S6K1 activation can occur (15, 22, 27, 28).
The Ral-mediated control of translation noted in the present study is also unusual in that input from the RalA oncogenic pathway negatively impacts translation, at least with regard to FLIPS mRNA. While the full range of Ral-mediated translational regulation remains to be determined, it seems unlikely that Ral pathway activation negatively regulates the translation of all mRNAs, particularly in light of recent studies showing that even complete disruption of PDK, an activator of Akt-mTOR signaling, suppresses the translation of only a subset of mRNAs and enhances the polysomal association of others (44). Rather, the Ral-mediated suppression of S6K1 activity may suppress the translation of only a subset of structurally related mRNAs. If this is the case, the examination of the differential sensitivity of the closely related FLIPS and FLIPL mRNAs to Ral-mediated translational regulation may prove informative. Alternatively, RalA may be linked to the translational apparatus by both an mTOR-independent pathway that suppresses the translation of FLIPS and an mTOR-dependent pathway that does not regulate FLIPS mRNA translation but does regulate the translation of other mRNAs critical for Ral-mediated transformation. It's worth noting that RalA appears to control not only FLIPS translation but also FLIPS stability, as RalA pathway activation significantly decreases the half-life of the FLIPS protein in a manner that is reversible by the overexpression of S6K1 (unpublished observations). RalA may therefore have multiple inputs into translational and posttranslational regulation and multiple means of controlling the levels of proteins critical for the translation process.
It is finally worth noting that the present studies also suggest that mTOR contributes in a unique manner to the control of the extrinsic apoptotic pathway by enhancing FLIPS localization to the DISC. Although mTOR has not been reported to be involved in protein trafficking or to interact with FLIPS, FLIPS is extensively posttranslationally modified in ways that could potentially be altered by mTOR (49, 50). Furthermore, because mTOR plays a critical role in Akt-mediated enhancement of FLIPS translation (31), the ability of mTOR to facilitate proper FLIPS localization may ensure proper shutdown of the extrinsic apoptotic pathway in tumors with activated Akt pathways. The ability of the Ral pathway to bypass this shutdown and to reopen the extrinsic cell death pathway independently of mTOR and even in the face of high Akt levels (unpublished data) suggests that therapies designed to block mTOR and the oncogenic stimulation of translation should be compatible and perhaps synergistic with approaches designed to activate the extrinsic pathway opened by RalA. It may also be possible that part of the effect of mTOR inhibition itself is the unmasking of apoptotic pathways stimulated by RalA, particularly if the Ral pathway suppresses the expression/translation of a broad set of antiapoptotic proteins.
The present studies define several unique properties of the Ral oncogenic pathway. RalA activation is linked to the translational machinery in a novel, mTOR-independent, negative manner. Furthermore, unlike other oncogenic pathways, RalA uses this link not to drive transformation but to sensitize cells to cell death. While the exact means by which RalA can cooperate with other Ras effectors to bring about transformation remain to be fully defined, the present studies provide a basis for understanding the linkage between Ral pathway activation and translation and a mechanistic understanding of how the extrinsic apoptotic pathway could be manipulated for therapeutic benefit in the multiple malignancies in which Ras and RalA activation play key roles.
This work was supported by National Institutes of Health Awards RO1 CA94989, RO1 CA115638, and P50 CA97257 to R.O.P.
†Supplemental material for this article may be found at http://mcb.asm.org/.
Published ahead of print on 7 August 2006.