|Home | About | Journals | Submit | Contact Us | Français|
Mammalian target of rapamycin (mTOR) is a nutrient and ATP sensor suggested to play an important role in tumorigenesis, particularly in the setting of PTEN loss or activated Akt/PKB. Of mTOR’s two known effectors, eIF4E has been implicated in tumorigenesis, while the role of S6Kinase (S6K1) in transformation is less understood. To assess the contribution of S6K1 to the transformed phenotype, we pharmacologically and genetically manipulated the mTOR-S6K pathway in glioma cells and monitored effects on growth in soft agar, a hallmark of cellular transformation, and also assessed in vivo intracranial growth. Anchorage-independent growth by HRasV12-transformed human astrocytes as well as by U251 and U373 human glioma cells was inhibited by pharmacologic mTOR inhibition. Similarly, shRNA-mediated suppression of mTOR also reduced anchorage-independent growth of glioma cell lines. Expression of wildtype eIF4E in rapamycin-treated E6/E7/hTert/HRasV12 and U373 cells failed to rescue colony formation, although expression of wildtype S6K1 or rapamycin-resistant S6K1 in rapamycin-treated U373 and U251 provided partial rescue. Consistent with the latter observation, siRNA-mediated suppression of S6K1 in HRasV12-transformed human astrocytes, U251, and U373 cells resulted in a significant loss of anchorage-independent growth. Furthermore, we found that in-vivo shRNA-mediated suppression of S6K1 in HRasV12-transformed human astrocytes reduced intracranial tumor size, in association with reduced tumor levels of phosphorylated ribosomal protein S6. These findings implicate the mTOR-S6K pathway as a critical mediator of glial cell transformation.
Mammalian Target of Rapamycin (mTOR) is an ATP and nutrient sensor that contributes to the control of cell size and cell cycle progression. mTOR’s ability to control cell size and cell cycle is due at least in part to its ability to regulate the translation of specific classes of mRNAs. mTOR-mediated control of translation is a rapamycin-sensitive process accomplished by regulation of the downstream targets, S6K and eIF4E (1–3). mTOR phosphorylates S6K, leading to phosphorylation of the ribosomal protein S6 and subsequent increased translation of mRNA with 5′ terminal oligopyrimidine sequences (2). In contrast, inhibition of eIF4E by the translational repressor 4EBP1 is reversed when 4EBP1 is phosphorylated by mTOR, resulting in the release of eIF4E, which can associate with eIF4A, eIF4G, and eIF4B to initiate the translation of capped mRNA (4).
The rapamycin-sensitive translational functions mediated by S6K and 4EBP1 have been recently recognized to be a result of mTOR’s interaction with raptor to form the mTORC1 complex, while rapamycin-insensitive functions are a result of mTOR’s interaction with rictor, forming mTORC2 (5–9). It remains to be determined how regulation of mTOR by raptor and rictor is coordinated, although each appears to control distinct and mutually exclusive mTOR functions. mTORC1, but not mTORC2, activates S6K, which can then inhibit insulin receptor substrate-I (IRS-1), thereby limiting insulin receptor mediated signaling through PI3K. mTORC2, in contrast, has recently been shown to phosphorylate PKB at Ser 473, thereby functioning as a PDK-2 (7).
Substantial indirect evidence indicates that mTOR fulfills a central role in tumor development and maintenance. Oncogenic signaling through a variety of molecules, such as the Epidermal Growth Factor Receptor, Ras, and PI3K, can upregulate mTOR activity and promote neoplastic growth (10). Tumors lacking normal Akt control mechanisms have also been shown to be particularly vulnerable to mTOR inhibition (11) and evidence of elevated mTOR activity can be found in multiple types of tumors (12), including malignant gliomas (13). These findings have led to the idea that mTOR plays a role in tumor maintenance, and to the development of mTOR inhibitors as systemic therapy against a wide range of malignancies.
Despite evidence of a link between mTOR, S6K and eIF4E in response to growth factor activation, it is unclear whether S6K and/or eIF4E connect mTOR to tumor development and growth. Evidence from model systems has implicated eIF4E and S6K in tumor development in specific oncogenic contexts. For example, overexpression of eIF4E has been shown to transform rat fibroblasts in cooperation with v-myc or E1A, and in vivo eIF4E overexpression causes the development of lymphomas, angiosarcomas, lung adenocarcinomas and hepatocellular adenomas (14–17). Inhibition of cap-binding by eIF4E also suppresses eIF4E-driven transformation (15). Although S6K has not been described as an oncoprotein, phosphorylated S6 protein levels are elevated in various tumor types, including malignant glioma (13, 18), and translational targets of S6K such as HIF1α appear to be critical in supporting tumor growth (19). Tumors with elevated HIF1 are sensitive to mTOR inhibition, and expression of HIF1α 5′ TOP sequences confers sensitivity to the mTOR inhibitor CCI-779 (20). Recent data also indicates that inhibition of angiogenesis by the tumor suppressor promyelocytic leukemia protein is in part dependent on its ability to inhibit mTOR and the synthesis of HIF1α (21). While these data suggest that eIF4E and S6K may directly mediate transformation through mTOR, amplification or mutation of eIF4E or S6K has not been found in spontaneously arising tumors, nor is mTOR itself thought to be an oncogene. Thus, the contribution of eIF4E and S6K to mTOR-dependent glial transformation remains open.
In order to test whether mTOR-dependent transformation requires both eIF4E and S6K functions, we genetically and pharmacologically manipulated mTOR and its downstream effectors and monitored effects on the transformation status of human glioma cell lines and transformed human astrocytes. We found that suppression of mTOR or raptor was sufficient to significantly reduce anchorage-independent growth in soft agar, an assay of transformation. Furthermore, S6K1, but not eIF4E, rescued glioma growth in soft agar from rapamycin-mediated suppression, and transient S6K1 inhibition was sufficient to significantly reduce glioma growth in soft agar. In vivo S6K1 suppression in intracranially implanted glioma xenografts reduced levels of phosphorylated S6 and also resulted in reduced intracranial tumor growth. This data is the first direct demonstration of S6K’s importance in supporting tumor growth both in vitro and in vivo. Collectively, these findings define a significant role for the mTOR-raptor (mTORC1)-S6K pathway in supporting gliomagenesis.
Human glioma cell lines were obtained from the Brain Tumor Research Center Tissue Bank at the University of California, San Francisco. Immortalized human astrocytes and HRasV12- and HRasV12/Akt-transformed human astrocytes (E6/E7/hTert, E6/E7/hTert/HRasV12 and E6/E7/hTert/HRasV12/Akt, respectively) were generated as previously described (22, 23). All cell lines were grown in Dulbecco’s modified Eagle medium (DMEM, 4500 mg/L glucose, Life Technologies, Inc.) supplemented with 10% fetal bovine serum (GIBCO-BRL, Invitrogen, Carlsbad, CA), penicillin, and streptomycin. Cells were grown in a humidified incubator containing 8% carbon dioxide at 37°C.
To assess cell proliferation, the 3-(4-5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt (MTS) assay was used (CellTiter96, Promega, Madison, WI). Cells were plated in triplicate into 96-well plates at a concentration of 2000 cells/well. (100 μl/well). At specified time points, 20 μL of MTS reagent were added to each well and allowed to incubate for one hour. Absorbance (490 nm) was then determined in a 96 well plate reader.
A pCAN1 vector encoding wt-eIF4E was a gift from Frank McCormick (University of California, San Francisco, San Francisco, CA). To generate a construct permitting wt-eIF4E expression with a unique selectable marker in E6/E7/hTert/HRasV12 and E6/E7/hTert/HRasV12/Akt cells, wt-eIF4E was subcloned from pCAN1-wt-eIF4E into the retroviral vector pMXI, which encodes green fluorescent protein (GFP) as a marker. Subcloning was performed as follows: pCAN1-wt-eIF4E was digested with XhoI and EcoRI, followed by gel purification of the wt-eIF4E encoding insert and ligation of the insert into identically digested pMXI. The retroviral pBABE constructs encoding wild type S6K1 or rapamycin-resistant S6K (pBABE/F5A-E389) were gifts from John Blenis (Harvard Medical School, Boston, MA). Retroviral vectors were used to infect cells as previously described (22). siRNA targeting S6K1 (Ambion #51221, Austin, Texas), 4EBP1 (Ambion #14089, Austin Texas) and control scramble sequence (Ambion #4611, Austin, Texas) were transfected using FUGENE 6 (Roche, Basel, Switzerland). Monolayer cells, grown to approximately 80% confluence, were exposed to retroviral supernatants with 8 μg/ml polybrene. Pools of productively infected cells (obtained by selection with puromycin, or hygromycin) were used for further analyses. Cells expressing GFP were separated by fluorescence activated cell sorting on a FACSVantage sorter (Becton-Dickinson, Franklin Lakes, NJ) located in the UCSF Laboratory for Cell Analysis at a band pass filter of 530/30 nm. Sorting gates were set such that 99% of the negative population and less than 1% of the positive population were excluded from the collection. Pooled collected cells were used for further analyses.
Lentiviral shRNAs targeting mTOR (6) was obtained from Addgene Inc. (Cambridge, MA). The lentivirus was packaged by co-transfection of 293T cells with the shRNA expression vector, VSV-G (vesicular stomatitis virus-glycoprotein), and delta-VPR plasmids at the ratio of 1:0.9:0.1, using FUGENE 6 (Roche, Basel, Switzerland). Forty eight hours after transfection, the supernatants containing lentiviral particles were harvested. Monolayer cells, grown to approximately 80% confluence, were exposed to the above lentiviral supernatants in the presence of 8 μg/ml polybrene for forty eight hours, followed by selection with 2 ug/ml puromycin for one week. After antibiotic selection, pools of productively infected cells were used for further analyses.
S6K1 was silenced in a tetracycline-inducible fashion by cloning a 97mer hairpin oligonucleotide targeting the S6K1 transcript into the pPRIME Tet-inducible construct (24). The following sequence was used to generate shRNA targeting S6K1: forward 5′ CCCCTGTCAGCCCAGTCAAATT TTCAAGAGA AATTTGACTGGGCTGACAG TTTTT -3′, reverse 3′ GGGGACAGTCGGGTCAGTTTAAAAGTTCTCTT TAAACTGACCCGACTTCAAAAA -5′. This sequence was cloned into pPRIME Tet-GFP-FF3 (a kind gift from Stephen J. Elledge, Harvard University) producing pPRIME Tet-GFP-S6K1 (SCT). The cloning product was confirmed by sequencing, virus was generated as described above, and 293T cells were infected with either pPRIME Tet-GFP-FF3 (targeting firefly luciferase), 8 μg VSV-G, 8 μg pCMV, and 16μg lentiviral vector, and used as a negative control) or pPRIME Tet-GFP-SCT. Viral supernatant was concentrated using Centricon Plus-20 filters (Millipore, Billerica, MA), then added to HRasV12-transformed human astrocytes. After four days of incubation, infected cells were sorted for GFP expression, and this GFP-positive pooled population was maintained in standard culture conditions described above. Doxycycline (Sigma-Aldrich, St. Louis, MO) at 5 μg/ml in standard media was added to induce expression of shRNA in culture.
As previously described (22), cells (1×104) were plated in DMEM plus 10% FCS in 0.35% (w/v) low melting temperature agar between layers of 0.7% low melting temperature agar. After 3 weeks, colonies were stained with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (Sigma-Aldrich, St. Louis, MO) and colonies of greater than fifty cells were scored by counting under a microscope. All experiments were performed in at least quadruplicate.
Immunodeficient rats (rnu/rnu; NCI, Bethesda, MD) were injected intracranially as described previously (22) with HRasV12-transformed human astrocytes expressing either control lentivirus (Ras Tet) or SCT (Ras SCT). Three days after injection, animals were fed either LabDiet 5053 or LabDiet 5053 (Purina, Richmond, IN) supplemented with Doxycycline at 6000 ppm daily. After fourteen days of exposure to doxycycline or control feed, animals were sacrificed, perfused with 4% paraformaldehyde, and brains were fixed in paraformaldehyde and paraffin-embedded.
Cells were harvested in lysis buffer (150 mM NaCl, 20 mM Tris-HCl (pH 7.5), 1% NP40, 1 mM EDTA, 1 mM EGTA, 1 mM sodium orthovanadate, and protease inhibitor mixture (Boehringer Mannheim Co., Ingelheim, Germany) at 4°C. Lysate was centrifuged (12,000 × g) for 10 minutes at 4°C to remove insoluble components. Protein was quantitated by the Bio-Rad Dc protein assay. Equal amounts of protein were separated on SDS-PAGE 12–16% gels, then transferred to Immobilon-P PVDF membrane (Millipore, Billerica, MA). The membrane was blocked with 5% nonfat dry milk in Tris-buffered saline containing 0.1% Tween 20. The membrane was then incubated with primary antibody in TBST, followed by secondary antibody linked to horseradish peroxidase diluted in TBST. ECL Detection System for Western Blot Analysis (Amersham, Buckinghamshire, UK) was used according to the manufacturer’s instructions for antibody detection. An AlphaImager 2000 Documentation and Analysis System (Multi Image light cabinet photodensitometer) was used to quantify the appropriate bands (Alpha Innotech Corporation, San Leandro, CA). Primary antibodies used were: anti-raptor (#4978), anti-mTOR (#2972), anti-S6K (#9202), anti-phospho p70S6K Thr389 (#9205), anti-phospho S6 Ser235/236 (#2211) (all Cell Signaling, Beverly, MA), anti-alpha tubulin (Santa Cruz Biotechnology, Santa Cruz, CA).
Five micrometer sections were obtained through paraffin-embedded brain tumors at intervals of 500 μm. Sections were stained with haematoxylin and eosin, and the maximum cross sectional dimension of tumor was measured under a microscope by an observer blinded to treatment. The maximal tumor area was calculated on each consecutive slide through the tumors and summed for each tumor. Phosphorylated S6 Ser 235/236 (Cell Signaling, Beverly, MA) was assessed on paraffin-embedded sections using immunofluorescent staining as described previously (25). Images were captured and merged using Openlab (Improvision, Waltham, MA).
Statistical analyses were performed using the GB-STAT statistical package (Dynamic Microsystems, Silver Spring, MD). Standard errors were calculated for each mean, and statistical differences between groups were determined by Student’s t-test or ANOVA followed by Newman-Keul post-hoc tests as indicated.
Anchorage-dependent growth by tumors displaying activated PKB/Akt signaling, such as those lacking PTEN, is sensitive to mTOR inhibition (11), although mTOR’s role in maintaining anchorage-independent growth is less well defined. We used the mTOR inhibitor rapamycin to assess whether the anchorage-independent growth of two human glioma cell lines and two transformed human astrocytic cell lines was mTOR dependent. Immortalized human astrocytes transformed by the expression of HRasV12 or HRasV12/Akt grew in soft agar as did the U251 and U373 glioblastoma cell lines. The addition of rapamycin in the agar, however, suppressed the colony forming ability of all cells (data not shown). The anchorage-independent growth of HRasV12/Akt transformed human astrocytes was also no more resistant to rapamycin than that of human astrocytes transformed by HRasV12 alone, indicating that activated Akt failed to rescue anchorage-independent growth from suppression by rapamycin.
To confirm these results we also determined whether specific suppression of mTOR altered the growth of glioma cell lines in soft agar. To do so we stably introduced lentivirus encoding shRNA targeting mTOR (shmTOR) in U251 and U373, then assessed levels of mTOR and the downstream effectors phospho p70S6K Thr 389 and phospho Akt Ser 473 by western blotting. As shown in Figure 1A, shRNA targeting mTOR (shmTOR) selectively suppressed mTOR levels in both cell lines, leading to decreased phospho p70S6K Thr 389 levels. Consistent with prior observations, we also observed a concomitant increase in phospho Akt Ser 473 (26). While suppression of mTOR in cell lines did not significantly alter the proliferation rates of cells in culture (Figure 1B), it did suppress the growth of cells in soft agar (relative to scramble control) to an extent comparable to the loss of growth observed with rapamycin exposure (Figure 1C). These results demonstrate that mTOR plays a key role in maintaining anchorage-independent growth and transformation in glioma cells.
Because eIF4E has been shown to play a role in the transformation of 3T3 cells and in leukemogenesis, we expressed wildtype eIF4E in HRasV12-transformed human astrocytes, HRasV12/Akt-transformed human astrocytes, and U373 cells, then assessed the ability of eIF4E to alter the effects of rapamycin on growth in soft agar. As shown in Figure 2A, retroviral infection of HRasV12, HRasV12/Akt-transformed human astrocytes, and U373 cells increased eIF4E expression relative to vector control cells. eIF4E expression in HRasV12/Akt-transformed human astrocytes significantly increased colony formation in the absence of rapamycin relative to vector control cells (Figure 2B). eIF4E expression did not, however, reverse rapamycin-mediated suppression of colony formation in soft agar (Figure 2B). In contrast to the effects of eIF4E expression in HRasV12-transformed human astrocytes, eIF4E expression in U373 did not alter baseline soft agar colony formation. However, as in the above cell lines, eIF4E expression in U373 failed to confer resistance to rapamycin exposure. These data indicate that while eIF4E introduction confers a growth advantage, it does not rescue cells from rapamycin-induced suppression of growth in soft agar.
eIF4E function is negatively regulated by 4EBP1, and we further substantiated the above finding by silencing 4EBP1 in HRasV12-transformed human astrocytes and U251 using siRNA (Figure 2C), then assessing the ability of 4EBP1 to alter the effects of rapamycin on growth in soft agar. Reduction of 4EBP1 levels in both cell lines failed to rescue cells from the effects of rapamycin in soft agar (Figure 2D), providing further support that eIF4E does not significantly support mTOR-driven growth by gliomas.
S6K has been shown to modulate translation of messages possessing 5′TOP sequences, but has not been implicated in tumorigenesis. To directly address the role of S6K in transformation, we expressed wildtype S6K, or a constitutively active, rapamycin-resistant mutant S6K (E389R) in U373 and U251, then assessed effects on growth in soft agar with and without rapamycin present. As shown in Figure 3A, introduction of a vector encoding wildtype or constitutively activated S6K (E389R) increased levels of S6K and phospho S6 Ser 235/236 two to three fold in both cell lines relative to empty vector cells (pBabe). Having confirmed protein expression in viral transfectants, pooled transfectants were grown in soft agar in the presence or absence of rapamycin, and colonies were counted after three weeks. As shown in Figure 3B, colony formation by U251 cells expressing wt S6K remained sensitive to rapamycin exposure, while colony formation by U251 cells expressing the mutant S6K E389R was resistant to the presence of rapamycin. In U373 cells, expression of either wt S6K or the mutant S6K E389R resulted in partial rescue of soft agar colony formation in the presence of rapamycin, as compared to the empty vector (Figure 3B). The reduced expression of wt S6K as compared to the expression of mutant S6K E389R in the U251 cells may explain the absence of rescue from rapamycin-mediated suppression of soft agar growth that was observed in the U251 as compared to the U373 cells, which in comparison had more comparable protein levels of the wt and mutant forms of S6K. To further assess S6K’s importance in maintaining a transformed state, we performed the converse experiment by transiently transfecting U373, U251 and HRasV12-transformed human astrocytes with siRNA targeting S6K1, plating cells twenty four hours after transfection into soft agar and monitoring colony formation three weeks later. As shown in Figure 4A, siRNA transfection produced an approximately 50–70% reduction of total S6K1 protein levels in all three cell lines by ninety six hours after transfection, and a 50–70% reduction in phospho S6 Ser 235/236 levels. Cells transfected with siRNA targeting S6K1 grew to confluence at a similar rate as cells transfected with scramble control (data not shown). Consistent with the idea that S6K1 maintains anchorage-independent growth, however, S6K1 suppression was associated with a significant loss of colony formation in soft agar by U373, U251, and HRasV12 transformed human astrocytes (Figure 4B).
Since transient S6K1 knockdown significantly reduced anchorage independent growth in all three cell lines tested, we sought to determine whether S6K1 knockdown in vivo could similarly inhibit tumor growth. However, concerns regarding the appropriateness of transient RNAi, as performed above, for in vivo studies of orthotopic xenografts led us to favor a more stable and inducible system for RNAi-based experiments. To confirm our in vitro findings, we constructed a lentiviral construct encoding shRNA targeting S6K1 in a tetracycline-inducible fashion and expressed either this construct or control vector into HRasV12-transformed human astrocytes. Pooled cells expressing the constructs were isolated by FACS for GFP expression, and tetracycline-inducible iS6K1 was confirmed in vitro (Figure 5A). HRasV12-transformed human astrocytes expressing either control vector (Ras Tet) or vector targeting S6K1 (Ras SCT) were injected intracranially into rats. Based on previously described formulations for doxycycline-impregnated rodent feed (27), mice were exposed to either control feed or feed containing doxycycline. After a fourteen-day exposure to either control feed or feed containing doxycycline, animals were sacrificed and brains were fixed, sectioned and stained with hematoxylin and eosin. We then assessed in vivo phosphorylated S6 Ser 235/236 levels by immunohistochemistry, and found that Ras SCT and Ras Tet tumors demonstrated comparable levels of cytoplasmic phosphorylated S6 in the presence of control feed (Figure 5B). While exposure to doxycycline failed to alter levels of cytoplasmic phosphorylated S6 in Ras Tet tumors, Ras SCT tumors exposed to doxycycline had comparatively reduced cytoplasmic phosphorylated S6 (Figure 5B). Animals injected with Ras SCT tumors and receiving doxycycline feed developed smaller tumors as compared to control animals (Figure 5C, p<0.05, Student’s t- test). Similar to the degree of reduced soft agar colony formation after iS6K shown in Figure 4B, we observed approximately 50% smaller tumors in the animals injected with Ras SCT tumor cells and receiving doxycycline feed relative to controls. These data indicate that HRasV12-transformed human astrocytes demonstrate significant cytoplasmic levels of phosphorylated S6 in vivo, and suggest that S6K1 knockdown occurred in a doxycycline-dependent manner. These results show that maintenance of S6K1 activity supports HRasV12-transformation in vivo, and that loss of S6K1 activity compromises tumor growth.
mTOR and its downstream effectors, eIF4E and S6K, have been implicated in cellular transformation, although their contribution to glial transformation remains undefined. In this work we show that growth of glioma cells in soft agar, a stringent assay for transformation, is blocked by down regulation of mTORC1 and that signaling through S6K, but not eIF4E, maintains glial transformation. We also show that in vivo suppression of S6K results in reduced intracranial glioma growth. These findings indicate that the S6K arm may have special significance in glial transformation.
Our data suggest that mTORC2 function is less significant in mTOR-dependent anchorage-independent growth for a few reasons. Rapamycin has been reported to have alternate effects on Akt phosphorylation; prolonged rapamycin exposure has been shown to inhibit assembly of mTORC2, thereby inhibiting Akt (28), and mTOR inhibition has also been described to induce insulin receptor substrate-1 (IRS-1), leading to Akt activation (26). In the human glioma cell lines U251 and U373, we found that suppression of mTOR resulted in a significant increase in phosphorylated Akt Ser473 as compared to cells expressing scramble control. Despite the increase in phosphorylated Akt Ser473, mTOR knockdown nonetheless significantly compromised these cells’ anchorage independent growth. These data suggest that in gliomas, mTORC2 is not the dominant arm supporting mTOR-dependent transformation, although it is possible that mTORC2 has effects on tumorigenesis that are Akt independent.
Our data also suggest that S6K and eIF4E have distinct roles in gliomagenesis. While prior findings have shown that eIF4E transforms rat fibroblasts, we found that eIF4E expression in U373 glioma cells and HRasV12- and HRasV12/Akt- transformed human astrocytes failed to restore anchorage-independent growth in the setting of mTOR inhibition. Silencing 4EBP1 also failed to rescue anchorage-independent growth from rapamycin-mediated suppression. eIF4E overexpression, however, increased colony formation in HRasV12/Akt- transformed human astrocytes, suggesting that eIF4E plays a positive role in transformation, and it is possible that eIF4E’s effects on transformation require other mTOR-dependent pathways such as S6K1. Another reason we cannot fully exclude a role by eIF4E in glial transformation is that superphysiologic levels of eIF4E beyond the two to fourfold increases generated in this study may be required for transformation. Malignant gliomas have been described immunohistochemically to express more eIF4E as compared to normal neuroglial cells (29), although the degree of eIF4E overexpression in malignant gliomas remains undefined.
Although S6K has not been shown to be a n oncoprotein, in the human gliomas assessed in this study, S6K appeared to be a key factor in maintaining anchorage-independent growth. The actions of S6K demonstrated in human astrocytes may indicate that S6K has differing roles in various tissue types: for example, S6K1 deletion blocks growth factor stimulated hypertrophy in muscle but not in neurons (30). S6K has numerous downstream targets, among them mRNAs with 5′ TOP sequences, the protein products of which are starting to be understood. The critical targets of S6K are not well defined, although the present data clearly suggest that these targets may be distinct from those influenced by eIF4E, and may represent better therapeutic targets. It should be noted that while the mTOR-S6K pathway appears to be critical for the growth of cells in soft agar, transformation of glial cells requires a series of events (p53 inactivation, Rb inactivation, telomerase reactivation, Ras activation) to which the mTOR-S6K pathway is merely a contributor (22, 31). This observation is consistent with the finding that supplying S6K to rapamycin-treated cells only partially rescues growth. The present findings suggest that S6K, but not eIF4E, plays a key, although not sufficient, role in glial transformation.
In addition to our data, which shows an important role for S6K1 in supporting gliomagenesis in vitro and in vivo, recently published data describe ribosomal S6Kinase 2 (RSK2) as supporting anchorage independent growth induced by tumor promoting agents such as Epidermal Growth Factor (EGF) and 12-O-tetradecanoylphorbol-13-acetate (TPA) (32). RSK2, a homologue of S6K1, is similarly activated by mitogens and inhibited by rapamycin (33). Although both RSK2 and S6K1 phosphorylate S6 in vivo, these kinases do not appear to be functionally redundant for a few reasons. S6K1 knockout mice have a small-body phenotype, despite the finding that mouse embryo fibroblasts from these animals show normal S6 phosphorylation in vivo, suggesting that RSK2 does not completely duplicate S6K1 functions (34). Comparisons of amino acid sequences and localization between the two S6 Kinases also suggest distinct functional differences (33, 35). It remains possible that S6K1 and RSK2 support tumor growth through similar mechanisms, and further studies defining the transformation-promoting effects common or specific to these kinases are needed.
Defining the role of the mTOR-S6K pathway in glial transformation may have an impact on the design and implementation of glioma therapies. Current targeted therapies are based on our knowledge of pathways thought to be critical for tumorigenesis and proliferation. This rationale has led to the clinical testing of signaling inhibitors such as Tarceva and CCI-779. Despite this mechanistic approach to drug development, these agents have shown only modest effects, and combinatorial strategies that inhibit multiple kinases (for example PI3K or Akt in combination with mTOR) show more promise than strategies employing single kinase inhibition (36, 37). In the case of Akt/mTOR combinatorial therapy, the fact that mTOR inhibition can induce Akt activation through IRS-1 may explain why targeting the same pathway at multiple sites is associated with better efficacy. Concerns have been raised that Akt activation with mTORC1 inhibition could represent a mechanism for drug resistance and sustained tumor growth, although in our model, Akt activation did not rescue tumor growth from mTORC1 inhibition. Our observation that the mTOR-S6K pathway plays a key role in glial transformation suggests that targeting the Akt-mTOR-S6K pathway at a more distal point may be as effective as dual inhibition at a more proximal point. Selective S6K inhibitors are not at present available at the clinical or pre-clinical level, although the present studies suggest that such agents, alone or in combination with other agents, might be rational choices for glioma therapy, and perhaps other tumors dependent on mTOR-S6K signaling for maintenance of the transformed phenotype.
American Brain Tumor Association fellowship (to J.L.N.), American Cancer Society Institutional Grant (to J.L.N), 1K08CA115476-01 (to J.L.N.), National Institutes of Health Award R01CA94989 (to R.O.P.).