PI3Kinase signaling in glioblastoma

EGFR is the first member of the ErbB (Her) family of RTKs. The two main EGFR ligands are EGF and TGF-α (transforming growth factor-α), among other ligands like beta-cellulin, epiregulin, heparin binding EGF (HB-EGF), and amphiregulin. Ligand binding to EGFR induces receptor phosphorylation, which in turn activates a complex downstream signaling network [28]. Downstream signaling through PI3K-PKB [20], PI3K-Rac-Rho, Ras-Raf-Mek-Erk and Jak-STAT influence proliferation [29], migration [30], invasion, resistance to apoptosis [31, 32], and tumor neo-vascularization [33, 34] (Fig. 1). Overexpression of EGFR has been found in many different tumor types including GBM, and has been consistently found to be correlated with a poor outcome [12, 35]. Genetic alterations like overexpression, small deletions or mutations can lead to oncogenic upregulation of the receptor [36]. In GBM, activation of EGFR is present in 40–60% of tumors [37]. The most frequent activating mutation is the EGF mutant receptor vIII (EGFRvIII) [36]. Amplification of EGFR gene in GBM leads to downstream activation of PI3K/PKB/mTOR/rpS6. Interestingly, it has been shown that inhibition of EGFR signaling correlates with decrease p-mTOR and p-rpS6 in cells wild-type for PTEN. In contrast, inhibition of EGFR signaling fails to affect p-mTOR or p-rpS6 in cells mutant for PTEN. Recently, a PKB independent pathway linking EGFR to mTOR through PKC (protein kinase C) was described indicating an additional drug target [38].

Members of the phosphatidylinositol 3-kinase (PI3K) family are lipid kinases involved in diverse signaling pathways that regulate proliferation, differentiation, migration, trafficking, and glucose homeostasis [39‐42]. They contain a p110 catalytic subunit that heterodimerizes with five distinct regulatory subunits (p85α, p55α, p50α, p85β, and p55γ). The p110 catalytic subunit includes an N-terminal p85 binding domain, a Ras binding domain, a C2 domain, a phosphatidylinositol kinase homology (PIK) domain, and a C-terminal catalytic domain. The PIK and catalytic domains of p110 are homologous to other protein kinase domains including mTOR, ATM (ataxia telangectasia mutated), ATR (ataxia teleangectasia rad3 related) and DNA-PK (DNA-dependent serine/threonine protein kinase). Mutations within the p110 subunit of PI3K, that are mainly gain of function mutations, have been identified mostly in exon 9 (helical domain) and in exon 20 (kinase domain) [43, 44]. In mammals, 8 distinct PI3K have so far been described. They are divided into classes I–III according to their substrate specificity, regulation and structure. Class I PI3Ks contain two subgroups, IA (p110α, p110β and p110δ) and IB (p110γ), which are activated by growth factor receptor tyrosine kinase (RTKs) and by G-protein-coupled receptors (GPCRs), respectively [42]. Class II PI3Ks consist of a single p110-like catalytic subunit that regulates membrane trafficking and receptor internalization [41, 45]. Class III PI3Ks has been found to regulate mTOR activity in response to availability of amino acids for the control cell growth [46‐48].

PI3K are activated by a wide range of upstream signals and phosphorylate the lipid phosphatidylinositol-4,5,-bisphosphate, generating phosphatidylinositol-3,4,5-tris-phosphate (PIP3) [41, 42]. The protein serine/threonine kinase PKB (also known as AKT) is the principal PIP3 target. PKB is recruited to the membrane upon binding of PIP3 with subsequent phosphorylation by the mTOR-rictor kinase complex and by PDK1 (3-phosphoinositide-dependent kinase). Activation of PKB, in turn, phosphorylates many target proteins which regulate cell metabolism, cell cycle and cell survival [49, 50], protein synthesis [47], cell polarity, cell motility [51], and vesicle sorting [52] (Fig. 1).

Therefore, the signaling components PI3K/PKB/mTOR are central regulators of cell proliferation, growth, differentiation, and survival. Moreover, PI3K regulates migration and invasion, mainly by the Rho family members Cdc42, Rac and Rho [53, 54]. These small GTPases dynamically remodel the actin cytoskeleton and give rise to filopodia, lamellipodia and stress fibers [55, 56].

PI3K dysregulation is observed in a variety of tumors [44, 57‐60] including GBM [44, 59], but also in other non-neoplastic human disorders [61]. For example, attenuated PI3K signaling downstream of the insulin receptor significantly contributes to the type-2 diabetes phenotype [62]. In human cancers, p110 of PI3K is frequently amplified [44]. This observation has been confirmed by another study that described PI3K p110 mutations in 15% of glioma samples, and 21% in pediatric and 17% in adult brain tumor samples [63]. In addition, elements of the PI3K signaling pathway are also frequently mutated, such as PTEN [64, 65] and the carboxyl-terminal modulator protein (CTMB) [66, 67]. Transgenic and knockout mouse models have confirmed the role of the PI3K-PBK pathway in tumorigenesis [68]. The TCGA group reported novel in frame deletions in the adaptor domain of PIK3CA. Interestingly in PI3KCA wild-type patients were found mutation in PIK3R1. These mutations clustered always around amino acid residues that disrupt C2-iSH2 interaction reducing the inhibitory effect of p85α on p110α [26].

PTEN (Phosphatase and tensin homologue) tumor suppressor gene encodes a phosphatase that catalyzes the dephosphorylation of phosphatidylinositol 3,4,5 triphosphate (PIP3), negatively regulating the activity of PI3K [8]. The PI3K/PTEN/PKB-pathway is of critical importance for angiogenesis, cell proliferation and survival [31, 50]. Homozygous Pten-knockout mice are embryonically lethal while heterozygous animals are viable but develop various tumors [69]. Loss of function mutations of PTEN are frequent in GBM and activate PKB in a similar way as mutations in PIK3CA [70]. Epigenetic gene silencing by promoter methylation also inactivates PTEN [71]. In animal models, haploinsufficiency was sufficient to promote tumorigenesis for certain tumor types, and progressive reduction of PTEN resulted in increasingly aggressive tumors [72, 73]. Transfection of PTEN resulted in reduced proliferation and induction of cell cycle arrest at G0/G1, accompanied by inactivation of PKB phosphorylation at Ser-473 [74]. In addition, exogenous PTEN expression induces astrocytic differentiation in the presence of the ECM [74] while neural stem cells self-renewal is negatively regulated by modulating GO-G1 cell cycle entry [75, 76]. PTEN expression sensitizes GBM cells to radiation, but not to chemotherapeutic drugs [77]. Epigenetic and genetic inactivation of PTEN is associated with shorter survival in GBM patients [8, 12, 78, 79].

P85β knockout mice develop hypoinsulinemia and hypoglycemia [80, 81]. Loss of all isoforms of PIK3R1 (including p50α and p55 α) resulted in perinatal lethality and caused a decrease in the expression and activity of class IA PI3K catalytic subunits. Heterozygous disruption of PIK3R1 improved insulin signaling and glucose homeostasis [82].

Mice carrying homozygous deletions for either p110α or p110β were found to be embryonic lethal [83, 84]. Inhibitors of p110α blocked insulin-stimulated phosphorylation of PKB, while inhibitors of p110β had no effect on insulin-stimulated phosphorylation. These results suggested that p110α has a key role in the PI3K-dependent insulin signaling [62]. Direct tumorigenic effect was proved in conditional and prostate-specific PTEN mutation: ablation of p110 blocked PTEN^−/−-induced tumor formation [85].

Different RTK inhibitors have been developed and studied in different cancer types including GBM, targeting EGFR, PI3K, mTOR and PKB.

Two different types of EGFR inhibitors have been developed: monoclonal antibodies (MoAbs) and small molecule inhibitors of EGFR tyrosine kinase activity, competing for the ATP binding site. The mechanism of receptor inhibition differs between the two types of drugs. MoAbs interfere with EGFR activation by blocking the extracellular ligand-binding domain. Protein kinase inhibitors (PKI) block the intracellular tyrosine kinase-mediated signaling pathways. Many different MoAbs have been developed with different affinity, specificity and negative regulatory effect. Cetuximab (IMC-225, Erbitux™; ImClone systems, Princeton, NJ, USA), a MoAb which binds EGFR with higher affinity than the natural ligands, has fairly recently been FDA-approved for treatment of patients with EGFR-positive metastatic colorectal cancer, and since December 2005 by Swissmedic for the treatment of patients with squamous cell carcinoma of the head and neck in combination with radiotherapy. Regarding GBM, the hR3 (TheraCIM; CIMYM Biosciences, ON, Canada), a human high affinity MoAb to EGFR that is now in phase I/II clinical trial (NCT00369252), has shown a partial response in a GBM study. The use of mAb 806 (Ludwig Institute, Victoria, Australia) that targets mutant EGFRvIII on glioma cell lines and mouse xenografts overexpressing EGFRvIII led to a dose-dependent growth inhibition [99]. Mab-806 is now in preclinical trial in an orthotopic murine glioma model, using EGFRvIII-positive U87MG cells [100]. Moreover, a combinatorial study, which uses mAb 806 and mAb 528, showed additive antitumor activity in human tumor xenografts [99]. The use of Cetuximab in tumor cell lines and a xenograft model induces apoptosis and inhibits angiogenesis [101, 102]. The small molecule inhibitors of EGFR are less specific than the MoAb and therefore the clinical effect was found to be less predictable. Nevertheless, these drugs have a low molecular weight, allowing better tumor penetration and can be administered orally. Gefitinib (ZD1839; AstraZeneca, Wilmington, DE, USA) and Erlotinib (OSI-774, CP-358, 774, Tarceva; OSI Pharmaceuticals, in collaboration with Genentech and Roche Pharmaceuticals) have been FDA-approved for different cancer types and new clinical trials in GBM patients are ongoing. Gefitinib showed only a limited activity in GBM patients [103] and colorectal cancer [104]. It did not improve survival as mono-therapy in patients with non-small cell lung cancer stage III nor when following chemotherapy and radiation therapy [105]. AEE-788 (LymphoSign), a potent EGFR and VEGF2 inhibitor, is currently in Phase I clinical trial (NCT00116376) for GBM. Erlotinib appears to be more effective against malignant glioma than gefinitib when comparing the radiographic response rate, but none of them have a clear impact on survival.

The limited effect of these PKI against EGFR raised the question whether drug efficiency could not be improved by a combinatorial strategy. In a pre-clinical GBM cell line model blocking EGFR activity with the protein kinase inhibitory drug AEE788, we found a 10–20% reduction of cell proliferation but only a limited pro-apoptotic effect in the minority of GBM cell lines [106]. Gleevec (PKI of the PDGF receptor developed by Novartis) that inhibits Bcr-Abl and also has an activity against PDGFR and c-kit combined with AEE788 (Novartis) and RAD001 (an analog of rapamycin that inhibits mTOR developed by Novartis) was only marginally effective. However, the combination of AEE788 and patupilone (a cell cycle inhibitor blocking polymerization of microtubuli, developed by Novartis) synergistically induced apoptosis in 50% of GBM cell lines [106]. A synergistic effect was also found when combining erlotinib with the PI-103 that led to an increased proliferation block when compared with monotherapy or a combination of erlotinib and rapamycin. Such combinations may be effective against EGFR-driven PTEN mutant tumors [107]. EGFRvIII expression in GBM promotes DNA-PKcs (DNA dependent protein kinas catalytic subunit) and DBS (DNA double-strand breaks) repair most probably through the augmented PI3K/Akt1 signaling. This mechanism suggests that blocking the DBS and DNA-PKcs pathway together with an EGFR inhibition could lead to better results [108]. Using the EGFR inhibitor gefinitib together with an HMG-CoA reductase induces another interesting synergism, irrespective of EGFRvIII and PTEN status [109].

Several compounds inhibiting PI3K have been developed, e.g., wortmannin [110], Ly294002, staurosporine, quercetin, demethoxyviridin and PI-103 among others. Wortmannin and demethoxyviridin are potent, irreversible, but non-selective inhibitors of all PI3K. LY-294002 and quercetin are reversible and potent PI3K inhibitors, but non-selective and also bind other members of the PI3K family [111]. Most protein kinase inhibitors block access of ATP to the ATP-binding pocket by distinct stoichiometric interactions [112]. Wortmannin has shown activity against tumor xenografts from melanoma [113], breast [114], colon [114], ovarian [114], and pancreatic cancer [115]. Treating EGFRvIII glioma cell lines with wortmannin significantly suppressed cell growth to a similar extent as EGFR inhibitors [116]. Wortmannin has also been shown to be an effective radiosensitizer of glioma cells [117], and displayed, when tested on a panel of GBM cell lines, a slight pro-apoptotic effect [106]. LY-294002 efficacy has been studied in glioma cell lines. It significantly reduced the level of PtdIns (3,4,5)P₃, inactivated phospho-PKB, and blocked cell proliferation in a dose-dependent manner [118]. Moreover, Ly294002 blocks p53 induction by inhibiting stabilization of p53. Nevertheless, this down-regulatory effect on p53 by PI3K inhibitors (attenuating p53-dependent cell death) [119] may not be so relevant for many cancers, since the p53 pathway is inactivated in most cancers.

A new series of PI3K inhibitors, which selectively target different PI3K isoforms, were synthesized. The small molecule inhibitor PI-103 showed unique activity against genetically different GBM cell lines. PI-103 selectively blocked p110α and mTOR complex at nanomolar concentration [120]. Only down-regulation of p110α specifically blocked glioma proliferation in vitro, but not p110β [120]. PI-103 was efficient, irrespective of the genetic status of cell lines (PTEN, p53 and EGFR). In vivo data show that PI-103 was effective and non-toxic in glioma xenograft models [120, 121]. Since mTOR inhibition may lead to PI3K activation, theoretically, blocking mTOR combined with a PI3K inhibitor may induce cell death. PI3K inhibitors like Ly294002 enhance apoptosis triggered by TRAIL or cytotoxics (vincristine, doxorubicine, etoposide, etc.) [122]. Recently, an interesting paper showed that HOXA9 transcription was activated in GBM leading to decreased apoptosis and increased proliferation. Transcriptional activation of HOXA cluster was blocked by PI3K inhibitor through an epigenetic mechanism involving histone H3k27 tri-methylation [123].

The serine/threonine kinase PKB is one of the enzymes regulated by PI3K activity by regulating PtdIns-3,4,5-P₃ levels. Its activated form down-regulates Bad, caspase-9, GSK-3beta and forkhead transcription factors, suppressing apoptosis and promoting cell survival [31, 124]. In addition, PKB increased levels of vascular endothelial growth factor (VEGF) under hypoxic conditions [125]. PKB is activated in about 80% of GBM [126]. Activated PKB together with mutant Ras was found to be tumorigenic in a murine glioma model [68]. In another model using astrocytes expressing E6/E7, hTERT (human telomerase reverse transcriptase) and Ras, upgrading of a low-grade to a high-grade tumor was achieved by addition of activated PKB [127]. The effect of PKB in the regulation of cell cycle, apoptosis and angiogenesis of GBM cell lines strongly suggests a role in the development of GBM. The two PKB inhibitors, KP-372-1 and KP-37-2, induced apoptosis in GBM cells [128].

The TOR family of proteins has multiple functions: regulation of mRNA transcription and translation in response to essential nutrients, membrane trafficking, protein degradation, organization of the actin cytoskeleton, and PKC signaling [129, 130]. There is a rapamycin-sensitive mTOR-complex (mTORC1) and a rapamycin-insensitive complex (mTORC2). mTOR activates S6K1 and EBP1. mTOR can also be up-regulated indirectly through activation of PI3K and PKB. Different mTOR inhibitors have been developed and are being assessed in clinical trials: the rapamycin prototype and three rapamycin derivatives RAD001 (everolimus), CCI-779 (temsirolimus), and AP23573. All mTOR inhibitors form a complex with the intracellular immunophilin FKBP12 and inhibit mTOR. The wide spectrum of rapamycin covers infectious, immunosuppressive, endothelial and neurodegenerative disease.

Rapamycin has been shown to have a growth inhibitory effect in several human and murine cancer cell lines both in vitro and in a xenograft model [131‐134]. Rapamycin induces a decrease in cyclin D1 expression and an increase in p27 levels, leading to a cell cycle block in late G1/S-phase [135]. Rapamycin has also been shown to induce cell death in a limited number of tumor models, although the molecular mechanism leading to apoptosis is not clear. One of the mechanisms may be that the downstream target of mTOR—S6K1—inactivates the pro-apoptotic molecule BAD [136, 137]. Recently, it has been shown that inhibition of mTOR/S6K1 triggers a negative feedback loop resulting in the activation of AKT signaling probably through a IGF-1R-dependent mechanism. Inhibiting mTOR also has an effect on the angiogenesis by blocking the translation of HIF-α [138]. Anti-angiogenic effects on endothelial cells have further been shown in a rapamycin-sensitive murine tumor model [139]. An interesting paper from Liu et al. showed that NVP-BEZ235, a novel dual PI3K/mammalian target of rapamycin (mTOR) inhibitor, was able to inhibit PI3K and mTOR signaling,and induce cell cycle arrest, down-regulation of VEGF and autophagy in gliomas [140]. Different studies combining mTOR inhibitors together with other anti-cancer compounds have been published with contradictory findings [106, 141, 142]; however, in vitro results may not predict clinical efficacy given the wide spectrum of rapamycin effects. Different clinical trials are now being performed for the rapamycin derivates CCI-779 (http://​clinicaltrials.​gov/​ct2/​results?​term=​CCI-779), RAD001 (http://​clinicaltrials.​gov/​ct2/​results?​term=​RAD001) and AP23573 (http://​clinicaltrials.​gov/​ct2/​results?​term=​AP23573) alone or in combination with other drugs. The most promising results have been described for renal cell carcinoma [143], endometrial cancer [144], and mantle cell lymphoma [145], but also giant cell astrocytoma in tuberous sclerosis patients. Phase II studies have been performed in recurrent GBM patients, with limited anti-tumor activity [146, 147]. Given the negative feedback loop between mTOR and PKB via IGF-1R, an mTOR inhibitor combined with an IGF-1R antibody/inhibitor may be a promising strategy to increase therapeutic efficacy [148].

Use of drug combinations is of particular interest, given the limited responses obtained in clinical trials using single drug approaches. In cases of an additive or synergistic effect, drug doses of each respective compound can be reduced, potentially paralleled by a reduction in side effects. We studied the impact of different drug combinations on survival and proliferation of GBM cell lines. Targeting EGFR with AEE788 and PDGFR with Gleevec and/or mTOR with RAD001, we found that single and combined applications did not significantly enhance background apoptosis rates [106]. However, the combination of AEE788 with the microtubuli inhibitor patupilone induced apoptosis in some 50% of cell lines, which was accompanied by simultaneous inactivation of both p-ERK and p-PKB. Asking whether down-regulation of p-ERK and p-PKB is critical for GBM cell survival, we directly blocked the PI3K/PKB and Ras/Raf/MEK/ERK pathways with both the PI3K inhibitor wortmannin and the MEK inhibitor UO126. This combination frequently killed GBM cells, supporting a model of an additive effect by targeting these two signaling pathways that results in a reduced threshold for the induction of apoptosis. We did not find any correlation between the sensitivity or resistance of GBM cells to apoptosis and their genetic status [106]. Simultaneous treatment with rapamycin and the EGFR inhibitor (EKI-785) resulted in synergistic anti-proliferative and pro-apoptotic effects [149]. At a molecular level, rapamycin alone decreased S6 phosphorylation, while EGFR inhibitor reduced phosphorylation of STAT3 (signal transducer and activator of transcription). Rapamycin alone increased phosphorylation of PKB and promoted the binding of the translational inhibitor eukaryotic initiation factor 4 binding protein (4EBP1) to the eukaryotic translational initiation factor 4E (eIF4E), which are blocked by EGFR inhibition [150]. Simultaneous blocking of multiple enzyme activities may reduce the effect of compensatory signaling which is one of the therapeutic limitations of using single agent therapy. However, suppressing cross-talk may not be accompanied by an enhanced therapeutic effect. Furthermore, combining the anti-malaria drug artesunate (ART) with the EGFR inhibitor OSI-774 resulted in an in vitro cytostatic effect which was most pronounced in a background of constitutively active EGFR [151]. Results of a Phase II study of Imatinib mesylate (Gleevec) plus hydroxyurea showed that this combination was well tolerated and associated with a modest clinical response in a subgroup of patients with recurrent GBM [91]. PI3K inhibitor (as LY294002 and wormannin) sensitizes GBM cells to apoptosis. This mechanism act both activating extrinsic apoptosis through TRAIL (tumor necrosis factor-related apoptosis-inducing ligand) and CD95 and the intrinsic mitochondrial apoptotic pathway [152].