Author contributions: J.S.: concept and design, collection and assembly of data, data analysis and interpretation and manuscript writing; K.-I.M.: data analysis and interpretation; A.S.: concept and design, data analysis and interpretation; K.T.: data analysis and interpretation; K. Suzuki: provision of study material or patients; Y.N.: provision of study material or patients and data analysis and interpretation; S.S.: provision of study material or patients and data analysis and interpretation; K. Sakurada: provision of study material or patients, data analysis and interpretation; T.K.: provision of study material or patients, data analysis and interpretation; A.T.: data analysis and interpretation; C.K.: concept and design, data analysis and interpretation, manuscript writing and final approval of manuscript. K.-I.M. and A.S. contributed equally to this article.
First published online in STEM CELLS EXPRESS September 20, 2010.
Disclosure of potential conflicts of interest is found at the end of this article.
The molecular signaling pathways orchestrating the biology of cancer stem-like cells (CSLCs), including glioblastoma, remain to be elucidated. We investigated in this study the role of the MEK/extracellular signal-regulated kinase (ERK) pathway in the control of self-renewal and tumorigenicity of glioblastoma CSLCs, particularly in relation to the PI3K/mTOR (mammalian target of rapamycin) pathway. Targeted inactivation of MEK alone using pharmacological inhibitors or siRNAs resulted in reduced sphere formation of both cell line- and patient-derived glioblastoma CSLCs, accompanied by their differentiation into neuronal and glial lineages. Interestingly, this effect of MEK inactivation was apparently augmented in the presence of NVP-BEZ235, a dual inhibitor of PI3K and mTOR. As a potential explanation for this observed synergy, we found that inactivation of either the MEK/ERK or PI3K/mTOR pathway triggered activation of the other, suggesting that there may be mutually inhibitory crosstalk between these two pathways. Significantly, inactivation of either pathway led to the reduced activation of p70S6K, and siRNA-mediated knockdown of p70S6K resulted in the activation of both pathways, which no longer maintained the cross-inhibitory relationship. Finally, combinational blockade of both pathways in glioblastoma CSLCs suppressed their tumorigenicity, whether transplanted subcutaneously or intracranially, more efficiently than blockade of either alone. Our findings suggest that there is p70S6K-mediated, cross-inhibitory regulation between the MEK/ERK and PI3K/mTOR pathways, in which each contribute to the maintenance of the self-renewal and tumorigenic capacity of glioblastoma CSLCs. Thus, combinational disruption of these pathways would be a rational and effective strategy in the treatment of glioblastoma. STEM CELLS 2010;28:1930–1939
Glioblastoma is a grade IV astrocytoma, as defined by the World Health Organization and is the most common and malignant type of central nervous system tumors . Despite the wide range of treatments, including surgery, radiotherapy, and chemotherapy, the majority of therapy eventually fails. The outcome is even worse for patients with glioblastoma, with which long-term survival is exceedingly rare . Because of disappointing results with conventional therapies, new approaches are needed to significantly improve patient outcome in this aggressive disease.
There is increasing evidence that a variety of cancers, including glioblastoma, may be driven by a component of tumor-initiating cells that retain stem cell-like properties. These cancer stem-like cells (CSLCs) have been defined on the basis of their ability to self-renew, undergoing divisions that allow the generation of more CSLCs, spawning differentiated progeny, which contributes to tumor cellular heterogeneity, and seeding tumors in animal hosts . The potent tumorigenic capacity of CSLCs, coupled with evidence of radioresistance and chemoresistance, suggests that they contribute to tumor maintenance and recurrence, and that targeting CSLCs may offer new avenues of therapeutic intervention [4, 5]. Two main strategies are currently being exploited to eradicate the CSLC pool: (a) chemotherapeutic regimens that specifically drive CSLCs into cell death and (b) driving CSLCs into differentiation, thereby depleting the tumor reservoir. It is thought that differentiated cells have limited proliferative potential, lose the capacity for self-renewal and impair the potential to initiate tumor formation of bulk tumors. Furthermore, chemotherapeutic agents more effectively targeted non-CSLCs than CSLCs .
It has been reported that bone morphogenetic protein (BMP) 4 can inhibit proliferation and induce the differentiation of glioblastoma CSLCs into predominantly resembling mature astrocytes in vitro, thereby reducing their tumorigenic potential in mice , but BMP-mediated astroglial differentiation is impaired in a subpopulation of brain cancer stem cells due to epigenetic silencing of the BMP type B receptor, which causes a differentiation block contributing to the pathogenesis of glioblastoma . Transforming growth factor (TGF)-β-sex-determining region of Y chromosome-related high mobility group box (Sox) 4-Sox2 signaling maintains the tumorigenicity of glioma-initiating cells independent of leukemia inhibitory factor (LIF) and signaling inhibition deprives these cells by promoting their differentiation . Also, TGF-β increases glioma-initiating cell self-renewal through the induction of LIF, although it is unknown whether inhibition of TGF-β signaling induces differentiation . Recently, knock down of transformation/transcription domain-associated protein (TRRAP) has induced glioma-initiating cell differentiation and suppressed their tumorigenicity in vivo . As effective methods for inducing differentiation in glioblastoma CSLCs are limited, other methods are required.
One reason why glioblastoma is extremely difficult to treat is its complex biology. Studies of human glioblastoma samples have uncovered a large number of genetic abnormalities, of which deregulation of signal transduction pathways is one of the most prominent [12, 13]. Disruption of signal transduction in glioblastoma occurs through overexpression or a gain-of-function mutation of receptor tyrosine kinases, such as epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor, and fibroblast growth factor receptor [14–17]. These abnormalities lead to constitutive activation of Ras/extracellular signal-regulated kinase (ERK), PI3K/Akt/mammalian target of rapamycin (mTOR), and other signal transduction pathways. Indeed, the activity of ERK is elevated in the vast majority of glioblastomas [18, 19], and Akt is elevated in the majority of examined glioblastomas [20, 21]. Although no activating mutations in the Akt and mTOR genes are seen in glioblastoma, Akt and mTOR are also associated with loss of function of PTEN (phosphatase and tensin homolog) , which antagonizes the function of PI3K. However, only a minority of patients responded to treatment when rapamycin and derivatives (mTOR inhibitor) or MEK inhibitors were tested extensively in patients with highly proliferative tumors such as glioblastoma [13, 23, 24], suggesting that single-agent treatment may be insufficient to control this tumor. Recently, several studies reported that a combination of activated Ras and Akt induced high-grade gliomas with histological features of human glioblastoma in mice, while neither activated Ras nor Akt alone was sufficient to induce glioblastoma formation [21, 25]. Also, a combination of Raf-1 and Akt was able to form glioblastoma in mice . Loss of PTEN cooperated with v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRas) activation to induce glioblastoma from neural progenitors in mice, while PTEN loss alone was insufficient to cause glioblastoma . These findings suggest that the combined effects of the Ras/ERK and PI3K/Akt/mTOR pathways may be important for the formation of glioblastoma in mice, pointing to the possibility that combined blockage of these two signaling pathways may be an essential component of successful glioblastoma treatment strategies; however, it is unknown whether ERK, alone or in combination with PI3K, affects the fate of glioblastoma CSLCs. In this study, we therefore examined the role of the MEK/ERK pathway, as well as its possible interaction with the PI3K/mTOR pathway, in the control of glioblastoma CSLCs.
MATERIALS AND METHODS
Reagents and Antibodies
UO126 was purchased from Sigma (St. Loius, MO, http://www.sigmaaldrich.com), SL327 was from Tocris (Ellisville, MO, http://www.tocris.com), NVP-BEZ235 was from Axon Medchem (Groningen, Netherlands, http://www.axonmedchem.com), rapamycin was from Calbiochem (Darmstadt, Germany, http://www.emdchemicals.com), epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF) were from Peprotech (Princeton, NJ, http://www.peprotech.com), anti-nestin (AB5922) was from Chemicon (Billerica, MA, http://www.millipore.com), anti-Bmi1 (05-637) was from Upstate, and anti-Sox2 (MAB2018), anti-glial fibrillary acidic protein (GFAP) (AF2594), and anti-βIII-tubulin (MAB1195) were from R&D (Minneapolis, MN, http://www.rndsystems.com). Anti-phospho-Akt (#4058), anti-Akt (9272), anti-phospho-p70S6K (#9205), anti-p70S6K (#2708), anti-phospho-4EBP1 (#2855), anti-4EBP1 (#9452), anti-ERK1/2 (#4695), anti-phospho-ERK1/2(#9106) and anti-mTOR (#2972) were from Cell Signaling Technology (Danvers, MA, http://www.cellsignal.com). Anti-Musashi (ab212628) was from Abcam (Cambridge, MA, http://www.abcam.com), horseradish peroxidase (HRP)-conjugated secondary antibodies for immunoblotting were from Jackson Immuno Research (West Grove, PA, http://www.jacksonimmuno.com), and Alexa 488 -conjugated secondary antibodies for immunocytochemistry were from Molecular Probes (Carlsbad, CA, http://www.invitrogen.com).
Stem Cell Culture Medium
Serum-free Dulbecco's modified Eagle's medium (DMEM)/F12 medium (GIBCO, Carlsbad, CA, http://www.invitrogen.com) supplemented with N2 (AR003; R&D) was used as the stem cell culture medium for A172 CSLCs and serum-free DMEM/F12 medium containing B27 (17504-044; GIBCO) was used as the stem cell culture medium for patient-derived glioblastoma CSLCs (SJ28P3 and #38).
A172 Cell Culture
A172 glioblastoma cells were obtained from the RIKEN Bioresource Center and maintained in DMEM supplemented with 10% fetal bovine serum (FBS) and penicillin/streptomycin. For selective propagation of A172 CSLCs, A172 cells were seeded in 35-mm dishes at a density of 3 × 105 cells/ml in the stem cell culture medium in the presence of 20 ng/ml of EGF and bFGF. Fresh EGF and bFGF were added every day. Under these culture conditions, cell aggregates known as spheres were formed within 3 days after seeding and were mechanically dissociated and reseeded at 3- to 4-day interval. As sphere cells were prone to undergo spontaneous differentiation and/or cell death during serial passage, we cultured A172 CSLCs by the monolayer culture method [28, 29]. For the monolayer culture of A172 CSLCs, dissociated cells from spheres cultured for 14 days in stem cell culture medium with EGF and bFGF were plated onto collagen-coated dishes. Monolayer-cultured A172 CSLCs were dissociated by Accutase (Sigma) and reseeded once every 6–7 days. After at least 8 weeks, A172 CSLCs were used for experiments. Differentiation of A172 CSLCs was induced by culturing in DMEM/F12 containing 10% FBS for 10 days. Characterization of the monolayer-cultured A172 CSLCs is shown in Supporting Information Figure 1.
Culture of Patient-Derived Glioblastoma CSLCs
Primary human glioblastoma cells (SJ28P3 and #38) were derived from surgical specimens obtained after informed consent from glioblastoma patients in accordance with a protocol approved by the Institutional Review Boards of the National Cancer Center and Yamagata University School of Medicine. Tumors were minced into small pieces with scissors, washed with saline and incubated in 0.05% pronase (53702; Calbiochem), 0.02% collagenase II (C-6885; Sigma), and 0.02% DNaseI (Sigma) in saline for 30 minutes at 37°C. Cells were washed once in saline, passed through a 70-μm cell strainer (Falcon, Franklin Lakes, NJ, http://www.bd.com) and cultured in the stem cell culture medium in the presence of 20 ng/ml of EGF and bFGF. Fresh EGF and bFGF were added every day. Tumor spheres were dissociated and reseeded once every 3–4 days. Cells were plated onto collagen-coated dishes (IWAKI, Tokyo, Japan, http://atg.ushop.jp) for the monolayer culture of tumor spheres after the fourth passage. Monolayer-cultured SJ28P3 and #38 CSLCs were dissociated by Accutase and reseeded once every 6–7 days. On the sixth or seventh passage, SJ28P3 and #38 CSLCs were used for experiments. Differentiation was induced as described earlier for A172 CSLCs. Characterization of these patient-derived glioblastoma CSLCs is shown in Supporting Information Figure 1.
Intracranial xenografts: monolayer-cultured A172, SJ28P3, or #38 CSLCs (1 × 104) in 10 μl DMEM/F12 medium were injected stereotactically into the right cerebral hemisphere of 5-week-old male BALB/cAJcl- nu/nu mice (CLEA Japan, Inc., Tokyo, Japan, http://www.clea-japan.com) at a depth of 3 mm. In these xenograft models, the injected CSLCs formed brain tumors that became fatal within 2 months at the longest. In contrast, mice injected with as many as 1 × 107 of A172, SJ28P3 or #38 CSLCs induced to undergo differentiation in the presence of serum remained alive during the observation period (at least 6 months) without any tumor burden (data not shown). All animal experiments were performed under a protocol approved by the Animal Research Committee of Yamagata University.
Results are expressed as the means ± standard deviations and were analyzed using the unpaired Student's t test, whereas mouse survival was evaluated by the Kaplan-Meier method and analyzed using the log-rank test.
Blocking the MEK/ERK Signaling Pathway Inhibits Self-Renewal and Induces Differentiation of Glioblastoma Cancer Stem-Like Cells
To test the possible role of the MEK/ERK signaling pathway in glioblastoma CSLCs, we first examined the effect of the MEK1/2 inhibitors UO126 and SL327 on the sphere formation capacity of A172 CSLCs. This assay has been widely applied to monitor the self-renewal ability of CSLCs. MEK1/2 inhibitors were treated immediately after plating at various concentrations. As shown in Figure 1A, MEK1/2 inhibitors resulted in marked inhibition of A172 primary sphere formation. Treatment of A172 CSLCs with 10 μM MEK1/2 inhibitors reduced the level of ERK phosphorylation at Thr202 and Tyr204, which are downstream targets of MEK1/2 (Fig. 1B). It has been shown that the MEK/ERK pathway regulates cell viability in some cell lines [30, 31], so we investigated whether it could also reduce cell viability. As shown in Supporting Information Figure 2a, UO126/SL327 treatment did not affect cellular survival at any concentration tested. Next, we examined the expression of the neural stem cell (NSC)/progenitor and differentiation markers. Inhibition of the MEK1/2 cascade reduced the expression of Nestin, Musashi, Bmi1, and Sox2 (NSC/progenitor markers) and increased that of βIII-tubulin (a neural marker; Fig. 1C). We could not detect the expression of GFAP (astrocyte marker) using this assay. To further confirm the above results, we used RNA interference (RNAi) experiments. Figure 1E shows that the introduction of siRNAs targeting the MEK1 and MEK2 genes, but not control siRNA, reduced the amounts of endogenous MEK1 and MEK2 proteins and inhibited ERK phosphorylation. Similar to UO126 and SL327, depletion of MEK1 and MEK2 impaired primary sphere formation (Fig. 1G). Expression of NSC/progenitor markers (Nestin, Bmi1, and Sox2) was lower, although that of Musashi was unchanged, and that of βIII-tubulin was higher than control siRNA (Fig. 1F). Thus, the decrease in the number of primary spheres indicates that there were fewer sphere-forming cells due to inhibition of MEK1/2
For the secondary sphere formation assay, primary spheres were dissociated to form a suspension of single cells, and then cultured in stem cell culture medium in the presence of EGF and bFGF at a density of 1,000 cells/0.2 ml per 96-well plate to allow the formation of secondary spheres. The number of secondary spheres corresponded to the number of sphere-producing cells within the primary sphere, which itself is the product of the proliferation of a single sphere-producing cell. Therefore, the number of secondary spheres represents the self-renewing cell divisions arising from the original sphere-producing cell. The number of secondary spheres formed in the absence of inhibitors decreased at a concentration of 10 μM of UO126 or SL327 (Fig. 1D) and also decreased with the depletion of MEK1 and MEK2 (Fig. 1H). Taken together, these findings suggest that MEK/ERK signaling is critical for maintaining A172 glioblastoma CSLCs.
Inhibition of Both PI3K/mTOR and MEK/ERK Signaling Enhances Differentiation of Glioblastoma CSLCs
Although MEK/ERK pathway inhibition inhibited the sphere formation of glioblastoma CSLCs, the inhibitory effect became less pronounced with sequential passage (Fig. 2B). Therefore, we wondered whether concomitant inhibition of the PI3K/mTOR pathway would be more effective than targeting the MEK/ERK pathway alone. To inhibit the PI3K/mTOR pathway, we used a dual PI3K/mTOR inhibitor, NVP-BEZ235, as this agent effectively inhibited the phosphorylation of Akt, p70S6K, and 4EBP1, the downstream targets of PI3K and mTOR, in glioblastoma CSLCs (see Fig. 3). To examine the combination effect of NVP-BEZ235 and UO126/SL327 on the fate of A172 CSLCs, we treated the cells with either one inhibitor alone or a combination of both and analyzed their sphere formation ability. A172 CSLCs treated with both NVP-BEZ235 and UO126 for 3 days showed more prominent neurite outgrowth than those treated with either inhibitor alone, while those treated with the vehicle (dimethyl sulfoxide, DMSO) efficiently formed primary spheres (Fig. 2A). Significantly, A172 CSLCs treated with both NVP-BEZ235 and UO126/SL327 had reduced ability to form secondary, tertiary, and quarterly spheres, whereas those treated with either alone recovered sphere-forming ability through sequential passages, indicating effective blocking of the self-renewal ability of A172 CSLCs by combination treatment (Fig. 2B). We next considered whether concomitant inhibition of the PI3K/mTOR pathway would also promote the differentiation of A172 CSLCs induced by MEK/ERK pathway inhibition. Treatment of the cells with NVP-BEZ235 alone resulted in a modest increase of βIII-tubulin expression, but the combination of NVP-BEZ235 and UO126/SL327 induced a higher level of βIII-tubulin expression (Fig. 2C). In addition, A172 CSLCs treated with both NVP-BEZ235 and UO126 retained the differentiated morphology even after 6 days of inhibitor treatment, whereas those treated with either inhibitor alone began to resume sphere formation (Supporting Information Fig. 3). Importantly, similar results were obtained when patient-derived glioblastoma CSLCs (SJ28P3) were used, in which the expression of not only βIII-tubulin, but also GFAP, was more efficiently induced by combined treatment with NVP-BEZ235 and UO126/SL327 (Fig. 2D–2F and Supporting Information Fig. 3). These results were also confirmed immunocytochemically (Supporting Information Fig. 4). In patient-derived glioblastoma CSLCs (SJ28P3), the proportion of cells positive for βIII-tubulin or GFAP increased partially with single inhibitor treatment but increased markedly with two-inhibitor treatment. We also immunolabeled A172 CSLCs, however, as ∼95% of the cells became immunopositive for βIII-tubulin even with single inhibitor treatment, we could not clearly observe any further increase of βIII-tubulin-positive cells with two-inhibitor treatment (data not shown). Cell viability analysis revealed that the combination of NVP-BEZ235 and UO126/SL327 had only a modest effect on cellular survival (Supporting Information Fig. 2b and c). These findings indicate that dual inhibition of PI3K/mTOR and MEK/ERK has a profound prodifferentiation effect on glioblastoma CSLCs.
To confirm that these agents effectively blocked the corresponding signaling pathway in glioblastoma CSLCs, we investigated the activity of target molecules by immunoblot analysis (Fig. 3). NVP-BEZ235 inhibited the phosphorylation of Akt at Ser473, which was a downstream target of PI3K, and p70S6K at Thr389 and 4EBP1 at Thr37/46, which were downstream targets of mTOR (Fig. 3A, 3B). Inhibition of MEK1/2 by UO126 or SL327 resulted in reduced ERK1/2 phosphorylation in A172 CSLCs, as shown in Figures 1B and 3A. Interestingly, NVP-BEZ235-treated cells showed the robust activation of ERK and UO126/SL327-treated cells led to the activation of Akt, while the combination of NVP-BEZ235 and UO126/SL327 suppressed not only the level of phospho-ERK but also phospho-Akt. Similar events were observed in patient-derived glioblastoma CSLCs (Fig. 3B), therefore, these results suggest that glioblastoma CSLCs are induced to undergo differentiation more efficiently when both signaling pathways are inhibited by the combination of PI3K/mTOR and MEK/ERK inhibitors.
We also asked whether inhibition of mTOR alone is sufficient to promote UO126/SL327-induced glioblastoma CSLC differentiation, as some studies demonstrated that simultaneous inhibition of mTOR and MEK/ERK resulted in substantially enhanced antitumor effects in prostate and breast cancer, although the cells used in those conditions were bulk cells and not cancer stem cells [32, 33]. We used rapamycin, an mTOR inhibitor, and found that the expression level of βIII-tubulin and GFAP was the same level whether or not rapamycin was used in combination with UO126/SL327 (Supporting Information Fig. 5a and d). The sphere number with combination treatment tended to be lower than with UO126/SL327-only treatment but not significantly (Supporting Information Fig. 5c and f). These results suggest that inhibition of mTOR alone may not augment UO126/SL327 effects on glioblastoma CSLCs. Notably, phospho-Akt was also increased by rapamycin and appeared to be more activated when rapamycin was used in combination with UO126 or SL327 (Supporting Information Fig. 5b and e). It could therefore be possible that this increased activation of Akt contributes to maintenance of the self-renewal capacity and undifferentiated state of glioblastoma CSLCs.
Cross-Inhibitory Regulation of the MEK/ERK and PI3K/mTOR Pathways via p70S6K
The results in Figure 3 suggest that the PI3K/mTOR and MEK/ERK signaling pathways may have mutually inhibitory crosstalk, however, the effect of MEK/ERK inhibition on the PI3K/Akt pathway has not been well-characterized. We therefore adopted a genetic approach in addition to pharmacological inhibitors to confirm MEK inhibition-induced Akt activation (Fig. 4A). When the amounts of endogenous MEK1 and MEK2 proteins were reduced by RNAi-mediated knockdown, as shown in Figure 1E, the level of phospho-Akt increased. We also observed that siRNAs of both MEK1 and MEK2 reduced the level of p70S6K phosphorylation. MEK inhibitors (UO126 and SL327) also reduced the phosphorylation of p70S6K and 4EBP1, which are downstream targets of mTOR, in A172 CSLCs as well as patient-derived glioblastoma CSLCs (see Fig. 3).
Based on previous reports that the MEK/ERK pathway could modulate mTOR signaling , that activation of p70S6K inhibits upstream (PI3K-Akt) signaling [35, 36], and that p70S6K is a key element of the S6K/PI3K negative-feedback loop , we then investigated the possibility that PI3K/Akt activation by MEK/ERK inhibition is mediated by p70S6K in glioblastoma CSLCs. First, we checked the effect of p70S6K siRNA on Akt activity (Fig. 4B). Introduction of siRNAs targeted against p70S6K reduced the amount of p70S6K protein and increased the level of phospho-Akt in A172 CSLCs. Significantly, under this condition, phospho-ERK was also upregulated. As p70S6K is one of the downstream targets of mTOR, we examined the effect of knockdown- or inhibitor (rapamycin)-mediated inactivation of mTOR on ERK activity (Fig. 4C, 4D). mTOR inactivation by siRNA and rapamycin reduced p70S6K phosphorylation and increased Akt phosphorylation. Again, under these conditions, phospho-ERK was upregulated. These results suggest that p70S6K represses both the MEK/ERK and PI3K/Akt signaling pathways. Next, we asked whether PI3K/Akt activation by MEK/ERK inhibition is dependent on mTOR-p70S6K. The expression of both mTOR and p70S6K was suppressed most efficiently after 2–3 days of siRNA transfection (data not shown); therefore, cells were analyzed for phospho-Akt expression after 1 day of UO126/SL327 treatment, which was started 2 days after siRNA transfection. As shown in Figure 4C, 4E, and 4G, UO126/SL327 did not further increase phospho-Akt from the phospho-Akt level increased by knockdown of mTOR or p70S6K in glioblastoma CSLCs. UO126 also did not increase phospho-Akt under the condition of mTOR activity inhibition by rapamycin in A172 and patient-derived glioblastoma CSLCs (Fig. 4D, 4F). These results clearly indicate that the mTOR/p70S6K pathway is required for MEK/ERK inhibition-induced PI3K-Akt activation in these CSLCs. Similarly, NVP-BEZ235-induced ERK activation was dependent on p70S6K (Fig. 4H, 4I). Taken together, our results suggest that p70S6K is a key mediator of both MEK/ERK inhibition-induced PI3K/Akt activation and PI3K/mTOR inhibition-induced MEK/ERK activation.
Improved Efficacy of Signaling Pathway Inhibition and CSLC Differentiation Through Modulation of the Inhibitor Treatment Protocol
We checked whether other patient-derived glioblastoma CSLCs (#38) showed the same results as SJ28P3. NVP-BEZ235 or SL327-treated cells showed reduced numbers of primary, secondary, and tertiary spheres and induced the expression of βIII-tubulin and GFAP, but no combinational effects were observed (Supporting Information Fig. 6a and b). In a single use, NVP-BEZ235 inhibited its targets' phosphorylation (Akt, p70S6K, and 4EBP1) and upregulated ERK phosphorylation, while SL327 suppressed and increased the phosphorylation of ERK and Akt, respectively. When p70S6K siRNA reduced the amount of p70S6K protein, the level of phospho-ERK and phospho-Akt increased (Supporting Information Fig. 6d), suggesting that there is a negative feedback loop in #38 CSLCs, as in SJ28P3 CSLCs; however, simultaneous treatment with NVP-BEZ235 and SL327 somewhat increased the level of ERK phosphorylation and did not suppress Akt phosphorylation in #38 CSLCs (Supporting Information Fig. 6c). We therefore tested whether we could achieve more efficient signaling pathway inhibition by modulating the drug treatment protocol. As a result, we found that initial treatment of glioblastoma CSLCs with NVP-BEZ235 followed by the addition of SL327 is of significant benefit in this regard. NVP-BEZ235 inhibited phospho-p70S6K and enhanced phospho-ERK after 1 day of treatment. Importantly, subsequent SL327 suppressed phospho-ERK and phospho-Akt (Supporting Information Fig. 6e). In this condition, the expression of βIII-tubulin and GFAP appeared to be further increased (Supporting Information Fig. 6f). After sequential treatment, the formation of primary and secondary spheres was more efficiently inhibited than with the single-agent treatment (Supporting Information Fig. 6g). These results demonstrated that the efficacy of signaling inhibition by inhibitors is at least partly dependent on the drug treatment schedule and that efficient inhibition of both the PI3K/mTOR and MEK/ERK pathways is the key to effective induction of glioblastoma CSLC differentiation.
NVP-BEZ235 and UO126/SL327-Treated Glioblastoma CSLCs Lose the Capacity to Form Tumors In Vivo
As differentiation induced by the two inhibitors depletes the pool of CSLCs, which have been defined as cells with tumorigenic capacity, we inferred that glioblastoma CSLCs treated with inhibitors would have limited capacity to form tumors in nude mice. To this end, we injected vehicle (DMSO)- or inhibitor(s)-treated glioblastoma CSLCs 1 × 104 intracranially into nude mice (Fig. 5A and Supporting Information Fig. 7a). After 4 weeks, mice were sacrificed and analyzed by H&E staining. All animals receiving DMSO-treated CSLCs developed large tumor masses. Whereas NVP-BEZ235 or SL327-treated CSLCs formed smaller tumor masses than DMSO when used as a single-agent, combination-treated CSLCs formed smaller tumors than with the single-agent treatment (Fig. 5A and Supporting Information Fig. 7a). Similar results were obtained in subcutaneous xenografts (Supporting Information Fig. 8a and b). Consistent with these results, the survival of mice injected with NVP-BEZ235- and SL327/UO126-pretreated CSLCs was significantly longer than that of mice injected with CSLCs pretreated with either NVP-BEZ235 or SL327/UO126 alone (Fig. 5B and Supporting Information Fig. 7b). Collectively, the results presented here highlight the effectiveness of combined inhibition of PI3K/mTOR and MEK/ERK.
It is considered that differentiated cells from CSLCs have limited proliferative potential; therefore, differentiation therapy for glioblastoma CSLCs is expected to be one way of depleting the CSLC pool [38, 39]. BMP2/4 treatment and TGF-β inhibitor treatment were reported to induce the differentiation of glioblastoma CSLCs [7, 9]. Knockdown of TRRAP is also reported to significantly increase differentiation ; however, effective methods for inducing differentiation in glioblastoma CSLCs are restricted because it is poorly understood how glioblastoma CSLCs maintain the self-renewal capacity and multipotency of differentiation. We have shown in this study that (a) the MEK/ERK pathway plays an important role in regulating glioblastoma CSLCs, (b) blocking of both the PI3K/mTOR and MEK/ERK signaling pathways acts in combination to promote the differentiation of glioblastoma CSLCs and to inhibit their tumorigenicity in mice, and (c) PI3K/mTOR and MEK/ERK pathways repress each other through a p70S6K-mediated negative feedback loop.
Two of the most important signaling cascades frequently deregulated in cancer are the Ras/MEK/ERK and PI3K/Akt/mTOR pathways [40, 41]. There has been no report showing that MEK/ERK signaling is necessary to maintain glioblastoma CSLCs, while we have recently found that the PI3K/mTOR signaling pathway is critical for the maintenance of glioblastoma CSLC properties . We demonstrated here that blocking MEK/ERK activity by inhibitors or siRNA promotes the differentiation of glioblastoma CSLCs and reduces their tumorigenicity. One of the most common alternations in glioblastoma is the amplification/overexpression of EGFR [43, 44], potentially leading to the activation of downstream effectors, such as the Ras/ERK pathway. Even after chemotherapy and radiation therapy, the activation of ERK with glioblastoma was observed, although independently of EGFR , suggesting that ERK activation may have an important role in recurrence, therefore, the MEK/ERK pathway is a potential therapeutic target. On the other hand, combined activation of the Ras and Akt signal pathways is important for glioblastoma formation, at least in mice, although there are several different theories regarding the cell origin of glioblastoma [21, 25]. A single-agent PI3K pathway inhibitor or ERK pathway inhibitor had limited efficacy in the treatment of patients [13, 23, 24]. These reports give rise to the possibility that combined inhibition of these pathways may be effective in the treatment of glioblastoma. Consistent with this idea, targeting PI3K/mTOR and MEK/ERK signaling induced differentiation and inhibited the tumorigenic potential of glioblastoma CSLCs. In addition to the combined activation of H-Ras and Akt, the loss of p53 in GFAP-positive cells in the hippocampus or subventricular zone caused the formation of glioblastoma multiform-like disease  and loss of PTEN in conjunction with loss of p53 maintained glioma stem/progenitor cells in an undifferentiated state . Blocking both PI3K/mTOR and MEK/ERK signaling plus p53 pathway reactivation may therefore bring an additional benefit.
Generally, a hallmark of signaling networks is the presence of multiple nodes with a feedback loop and crosstalk between pathways in cells and it is important to clarify the molecular nodes. It has also been reported that the ERK pathway regulates the activation of mTORC1 through TCS2 phosphorylation . Although it is unclear whether ERK directly regulated TSC2 in our study, PI3K/Akt pathway activation by inhibiting MEK/ERK may rely on blocking of insulin receptor substrate by p70S6K or on blocking of rapamycin-insensitive companion of mTOR, which is an essential subunit of mTORC2, by p70S6K . In either case, it is possible that p70S6K plays an essential role in inhibiting both the PI3K/mTOR and MEK/ERK signaling pathways. Currently, the detailed mechanism underlying the cooperative differentiation effects of combined suppression of PI3K/mTOR and MEK/ERK activities in glioblastoma CSLCs remains unknown, and elucidation of this mechanism is an important future issue. Finally, our findings have implications for cancer therapy because they provide further explanations for the limited benefit observed in clinical trials and exemplify how effective PI3K/mTOR and MEK/ERK inhibition could be in the treatment of human cancer.
This work was supported by Grants-in-Aid for Scientific Research, Challenging Exploratory Research, Young Scientists, and for Scientific Research on Priority areas from the Ministry of Education, Culture, Sports, Science and Technology of Japan, by a Grant-in-Aid from the Global COE program of the Japan Society for the Promotion of Science, by a Grant-in-Aid for Cancer Research from the Ministry of Health, Labor, and Welfare of Japan, and by a grant from the Japan Brain Foundation.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.