GBM is poorly differentiated and resistant to traditional chemotherapeutic and radiotherapeutic approaches. The current standard treatment for GBM includes surgical resection of the primary tumor followed by radiotherapy and chemotherapy with temozolomide (TMZ) [1,2]. However, despite this aggressive treatment regimen, the prognosis for this disease remains poor, with a median overall survival of less than 15 months [3-5]. Recent evidence indicates that GBMs include a population of glioma stem cells (GSCs) that are responsible for the initiation, propagation, and recurrence of gliomas, and thus represent a major target of cancer therapy [6-9].

Although GSCs share properties with NSCs, subtle differences reside at both genetic and epigenetic levels, which lead to immortal growth, a block in differentiation and invasiveness.

Verhaak et al. clustered 1,740 genes expression data from 202 samples and divided GBM into four subtypes-proneural, neural, classical and mesenchymal--with 210 gene signatures for each subtype [10]. PDGFRA abnormalities, isocitrate dehydrogenase 1 (IDH1) and TP53 mutations are enriched in Proneural subtype. Mesenchymal subtype is associated with deletions of NF1. Amplification EGFR was observed in 97% of the Classical subtype. These signatures identified in different subtypes could be regarded either as biomarkers for prognosis or as potential therapeutic targets. This clustering is close to the previously described molecular subclasses of high-grade glioma [11]. Whether these GBM subgroups origin from the same initiating cell like previously proposed neural stem cell [12] or multiple stem cell-like populations remains to be established.

Recently, studies have identified several critical signaling pathways, including TGF-β, Notch, PI3K, activated in GSCs and are therefore promising therapeutic targets for GBM (Figure 1). Novel molecular strategies have been designed both in preclinical and clinical settings to target and interfere with those pathways with the hope to achieve longer patient survival. This review will highlight some of the signaling pathways activated in GSCs and important to the survival and proliferation of GSCs with emphasis on possible strategies for targeting these pathways for GBM therapy.

Figure 1: Schematic representation of the PI3K, MEK, Notch and TGFβ pathway with their inhibitors. TGF-β binds to and activates heterotetrameric TGF-β receptors, and leads to activation of Smad2 and Smad3. The phosphorylated Smad2/Smad3 complex interacts with Smad4 to form a heteromeric complex that translocates to the nucleus and regulates the transcription of a diverse array of genes. Both MEK-ERK signaling and PI3K signaling are activated by RTK. Notch is cleaved by γ-secretase and releases Notch intracellular domain (NICD). The NICD translocates to the nucleus, interacts with the CSL, and then converts CSL from a transcriptional repressor to an activator.

Notch signaling is a conserved pathway that plays a critical role in multiple cellular processes involved in cancer and development, including cell proliferation, survival, angiogenesis, migration, differentiation, and cell fate decisions [13,14].

Four Notch receptors (Notch1-4) and 5 Notch ligands (Jagged1 and Jagged2 [JAG1 and JAG2] and Delta-like 1, 3, and 4 [DLL-1, 3, and 4]) have been identified in mammals. Notch receptors are transmembrane receptors that contain both intracellular and extracellular regions. The extracellular regions of Notch receptors contain epidermal growth factor (EGF)-like repeats that are essential for ligand binding and juxtamembrane repeat motifs known as Lin-12/Notch repeats [15]. Notch signaling requires several proteolytic cleavages before activation. The first cleavage (at site S1) is regulated by a furin-like convertase [16,17]. Once the Notch receptor is activated by its ligand, cleavage occurs at the second proteolytic cleavage site (S2) of the Notch extracellular region by ADAM metallopeptidase domain 17 (ADAM17), also known as tumor necrosis factor-α-converting enzyme, an enzyme that belongs to the ADAM protein family of disintegrins and metalloproteases. Notch ligand binding can also induce a third proteolytic cleavage at the S3 site of the transmembrane region by γ-secretase; this cleavage releases the Notch intracellular domain (NICD) [18]. The NICD translocates to the nucleus, interacts with the DNA-binding protein Cbf1/Suppressor of hairless/Lag-1 (CSL), promotes dissociation of corepressor complexes from CSL, and then converts CSL from a transcriptional repressor to an activator [19,20].

Previous studies indicate that Notch signaling contributes to GBM progression. Kanamori et al. found that components of Notch signaling pathways were overexpressed in 51% of 35 primary GBMs, and genetic or pharmacological inhibition of Notch signaling selectively inhibited growth in cells exhibiting Notch pathway deregulation [21]. Consistently, Zhang et al. also reported that activation of Notch signaling promotes cell proliferation and colony formation in the human GBM cell line SHG-44 [22]. In addition, Notch cross talk with other signaling pathways promote GBM progression. For example, Notch1 promotes the invasive and migratory properties of GBM cells by stimulating β-catenin and NF- κB signaling [23]. Notch1 also mediates GBM cell proliferation and survival through the Akt-mammalian target of rapamycin (mTOR) signaling axis [24]. Blocking Notch pathway by γ-secretase inhibitors was reported to deplete CD133+ GBM cells, reduce neurosphere growth, and inhibit xenograft tumor growth through decreased Akt and STAT3 phosphorylation [25]. Combination of the Notch inhibitor MRK003 and the Akt inhibitor MK-2206 effectively inhibited GBM invasiveness [26].

In GSCs, the Notch signaling pathway plays an important role in cell proliferation. SiRNA targeting Notch-1 attenuated GSC proliferation in vitro and tumor growth in vivo [27]. Notch signaling pathway showed critical role in GSC maintenance and presented GBM subtype specificity. Saito et al. used several γ-secretase inhibitors (DAPT, BMS-708163, and RO4929097) to investigate the effects of Notch pathway inhibition in 16 GSC cell lines [15]. Interestingly, only proneural GSCs were sensitive to γ-secretase inhibitors. Further study found that 17 genes representing active Notch signaling components including NOTCH1, NOTCH3, hairy and enhancer of split 1 (HES1), mastermind-like 1 (MAML1), DLL-3, and JAG2 were highly expressed in proneural GSCs. Analysis of the Cancer Genome Atlas expression data identified a large percentage (43.9%) of GBM tumors with proneural expression signatures showing high Notch pathway activation; these findings suggest that γ-secretase inhibitors might be used for treating that particular GBM subtype [15]. Moreover, γ- secretase inhibitors impaired GSC maintenance and induced GSC differentiation [15]. These data suggest that pharmacological inhibition of the Notch signaling pathway may substantially improve GBM patient outcomes, particularly in patients with the proneural GBM subtype.

In addition, Notch signaling pathway was related with resistance to traditional GBM therapy. For example, Notch signaling promotes GSC radioresistance as γ-secretase inhibitors rendered GSCs more sensitive to radiation, and knockdown of Notch1 or Notch2 also resensitized GSCs to radiation and inhibited xenograft tumor formation [28]. Inhibition of the Notch signaling pathway also increased the sensitivity of CD133+ GSCs to TMZ [29]. These two studies suggested a cross talk between Notch signaling and DNA damage signaling, although the underlying mechanisms remain to be further elucidated.

Phosphatidylinositol 3-kinase (PI3K) pathway plays extensive role in multiple cellular functions such as cell proliferation, differentiation, motility, survival, and metabolism, and is one of the most commonly activated signaling pathway in human cancers [30,31]. This pathway therefore presents both an opportunity and a challenge for cancer therapy.

PI3K is activated by receptor tyrosine kinases (RTKs), such as EGFR. PI3K binds to phosphotyrosine residues on activated RTKs, and induces a conformational change of p85, then releases the inhibition from the catalytic subunit p110. Activated PI3K phosphorylates phosphatidylinositol 4,5-bisphosphate (PIP2) to phosphatidylinositol 3,4,5-trisphosphate (PIP3), which further recruits Akt to cell membrane and Akt is then phosphorylated by PDK1 (3- phosphoinositide-dependent kinase) and the mammalian target of rapamycin complex 2 (mTORC2). After activation, AKT is recruited to the cellular membrane and, in turn, phosphorylates, activates, or inhibits numerous target proteins involved in cell growth, proliferation, motility, and survival [30,32].

EGFR is amplified in approximately 45% of GBM cases and in 97% of the classical GBM [10,33]. PIK3CA (the p110a subunit of PI3K) or PIK3R1 (the p85 regulatory subunit of PI3K) is amplified or mutated in approximately 15% of GBM cases [34-36]. PTEN, the negative regulator of PI3K pathway, is inactivated by mutations, chromosomal deletions, or epigenetic gene silencing in about 40% of GBM cases [37]. Combined together, the aberrant activation of PI3K pathway by alterations of EGFR, PTEN, PIK3CA or PI3KR1 genes have been detected in over 63% of primary and 31% of secondary GBM [36-38] highlighting the critical role of PI3K pathway in GBM.

Recently, Brennan group reported that independent focal amplification of two or more RTKs (most commonly PDGFRA and EGFR) was identified in 34 of 463 cases in The Cancer Genome Atlas (TCGA) GBM dataset [39]. More interestingly, dual-color fluorescence in situ hybridization (FISH) performed on eight samples with EGFR and PDGFRA amplification revealed distinct tumor cell subpopulations amplified for only one RTK [39]. This important discovery may partially explain the failure of clinical trials of inhibitors targeting EGFR and PDGFR in GBM. Simultaneous targeting both EGFR and PDGFR would be necessary for inhibiting PI3 kinase pathway in GBM with intratumoral heterogeneity.

Because of the aberrant hyperactivation of the PI3K pathway in GBM and its pivotal role for tumor progression, inhibition of this pathway components is an attractive target for GBM therapeutics. Many of the PI3K inhibitors currently in pre- or clinical development inhibit all of the catalytic subunit isoforms of class IA PI3Ks (p110a, p110b, and p110g). For example, BKM120 (Novartis), a pan-class I PI3K inhibitor, causes dose-dependent growth inhibition in GBM cell lines (including U87, U251, LN229, LN18, and D54) regardless of their phosphatase and tensin homolog (PTEN) and/or epidermal growth factor receptor (EGFR) status. Furthermore, when given orally to mice with intracerebral tumor xenografts, BKM120 prolonged their survival and was well tolerated [40]. Because BKM120 can cross the bloodbrain barrier, it is an attractive option for the treatment of GBM. Similarly, NVP-BEZ235, a dual PI3K/mTOR inhibitor, has antitumor activity in human GBM [41]. Several other PI3K inhibitors are being developed and evaluated [42,43].

Multiple studies have demonstrated that the PI3K/Akt pathway not only promotes GBM progression but also contributes chemoresistance. For example, activation of PI3K signaling supports GBM migration and survival [44] and also enhances chemoresistance to TMZ [45]. Chen et al. found that blockade of the PI3K/Akt pathway by LY294002 enhances the cytotoxicity of TMZ in GBM cells [46]. The PI3K/mTOR dual inhibitor NVP-BEZ235 increased the radiosensitivity of GSCs in vitro by inducing apoptosis and by increasing levels of BID, Bax, and active caspase-3, and decreased level of Bcl-2 [47].

Besides, PI3K/Akt pathway is also activated by and cross talks to other signaling pathways. For example, CD133 phosphorylation mediates CD133’s interaction with the p85 subunit of PI3K, activates the PI3K/Akt pathway, and promotes the tumorigenicity of GSCs [48]. ADAM17 induces CD133+ GSC invasion and migration through EGFR/PI3K/Akt pathway activation [49]. The cross talk between PI3K/mTOR pathway and the MEK/ERK pathway was reported to be important for the maintenance of GSCs [50]. When the PI3K pathway is inhibited by BKM120 or NVP-BEZ235, ERK signaling is activated in GSCs, this indicates that ERK signaling may be a mechanism of resistance to PI3K inhibition [51]. Consequently, a combination of NVP-BEZ235 and PD-325901 (an ERK inhibitor) is more effective at inhibiting GSC proliferation than either treatment alone [51]. Taken together, these data highlight the importance of the cross talk of PI3K signaling pathway with other signaling pathways in GBM, and combinational targeting the PI3K pathway with related pathways may be an effective therapeutic strategy in GBM.

Transforming growth factor beta (TGF-β) belongs to a family of cytokines that includes the bone morphogenic proteins, activins, and nodals, all of which regulate embryonic development and tissue homeostasis [52]. TGF-β binds to and activates heterotetrameric transmembrane serine/threonine kinase receptors composed of the type I and II TGF-β receptors (TβRI and TβRII) [53]. Activated TβRI phosphorylates Smad2 and Smad3, which then interact with Smad4 to form a heteromeric complex that translocates to the nucleus. The Smad complex interacts directly with Smad-binding elements and associates with other transcription factors, coactivators, or corepressors, thus regulating the transcription of a diverse array of genes [54].

Disruption of the TGF-β pathway has been implicated in GBM progression via promotion of cell proliferation, angiogenesis, tumor invasion, metastasis, and immune suppression [55]. Zhang et al. found that the TGF-β expression level was positively correlated with microvascular density and angiogenesis in GBM [56]. Consistently, another study also showed TGF-β promoted angiogenesis via the c-Jun N-terminal kinase pathway and macrophage infiltration in a zebrafish GBM xenograft model [57]. Lu et al. showed that TGF-β can promote invasion and migration of GBM cells through activation of ADAM17 [58]. In addition, another study from the same group indicated that TGF-β-induced microRNA 10a/10b expression promotes GBM cell migration through the suppression of PTEN [59] suggesting a cross talk between TGF-β between PI3K/pTEN/Akt pathway.

A number of studies have shown that TGF-β plays a crucial role in the regulation of GSCs. Ye et al. showed that TGF-β1 can increase the invasiveness of human GBM-derived CD133+ GSCs [60]. Peñuelas et al. found that TGF-β promotes self-renewal of GSCs through Smaddependent up regulation of leukemia inhibitory factor (LIF) and activation of the Janus kinase (JAK)-STAT pathway [61]. Beside the Smad-dependent pathway, TGF-β also cross talk with other pathway and activated non-Smad signaling to regulate GSCs. Wang et al. demonstrated that TGF-β and interleukin-1β cooperated to induce up regulation of a subset of stemness factor proteins (Bmi-1, Notch2, nestin, and LIF) and induced a GSC phenotype in vivo [62]. In addition, autocrine TGF-β signaling was reported to play an important role in maintaining the tumorigenicity of GSCs through the TGF-β/ Sox4/Sox2 pathway [63]. These findings suggest that inhibition of TGF-β signaling may inhibit the capacity of GSCs to initiate brain tumors. Indeed, TGF-β receptor inhibitors decreased the CD44(high)/ Id1(high) GSC population through the repression of DNA-binding proteins and the transcriptional regulators Id1 and Id3 [64], and treatment with LY364947, a selective ATP-competitive inhibitor of TGF-βRI, reversed GSC radiation resistance and improve therapeutic response in patients [65].

GBM is a very complex tumor composed of a heterogeneous mix of cells including GSCs. GBM recurrence is attributed to GSC resistance to conventional radiation therapy and chemotherapy. Therefore, current treatments targeting the bulk of the tumor are insufficient. However, therapies that directly target GSC also failed in clinical trials. There are several possibilities: 1) crosstalk and compensation between different signaling pathways: complimentary or collateral pathways are activated upon targeting one specific signaling and used by GSCs as predominant mechanism of escape. A study by Sunayama et al. showed that suppression of either the MEK/ERK or PI3K/mTOR pathway in glioblastoma stem-like cells induced activation of the other collateral pathway [50].

The study also indicated mutually inhibitory crosstalk between these two pathways, giving a clue that and combinational disruption of these pathways would be a rational and effective strategy for GBM therapy [50]. 2) GBM is characterized by a high degree of intratumoral heterogeneity. Amplification of multiple RTKs, either co-amplification or independent focal amplification of two or more RTKs (most commonly PDGFRα and EGFR), was identified within individual GBMs and simultaneous inhibition of both EGFR and PDGFR was necessary for disruption of PI3 kinase pathway activity in the mixed population [39]. 3) In gliomas, it is hypothesized that some differentiated tumor cells have the ability to revert to stem cell-like cells [66,67]. In this case, combination of radiation/chemotherapy against the bulk tumor together with targeting signaling pathways aberrantly activated and required for the function of GSC population would be more reasonable and effective treatment.

Additional attention should also be paid to interaction between tumor stem cells and the microenvironment (endothelial cells, extracellular matrix, cytokines, nitric oxide, oxygen levels) which promote self-renewal of GSCs and might influence the response of tumor stem cells to targeting therapies [68,69]. Therefore, an optimal therapeutic strategy should be directed against GSCs functions, in combination with therapies possibly affecting microenvironmental factors that sustain GSCs.

Nevertheless, recent studies has made tremendous progress in understanding the signaling pathways involved in GSCs, and various approaches towards targeted therapy have been tested (Table 1). As discussed above, results in preclinical models targeting PI3K, Notch, TGF-β signaling alone or in combination with traditional therapy (radiotherapy or TMZ) were proved to be effective to eliminate GSCs and inhibit GBM progression. Further in vitro and in vivo experiments as well as clinical trials are warranted to assess the success of these approaches. To identify which patient populations will benefit from these inhibitors and to optimize the combination of inhibitors of two or more pathways, as well as combination of targeting specific pathway with conventional radiation therapy, chemotherapy, and antiangiogenic therapies based on genetic and molecular characteristics of the tumors would be challenges but also promising to improve GBM therapy.

Table 1: Clinical trials of targeted therapies in GBM.