Integrins are heterodimeric cell surface receptors that mediate cell adhesion to the extracellular matrix (ECM) and support cell–cell interactions in a multitude of physiological and pathological situations. They are at least 24 known αβ heterodimers formed by a combination of 18 α and 8 β subunits bound non-covalently. Natural ligands of integrins are component of the ECM such as vitronectin, collagen, or fibronectin. Each αβ integrin pair has a defined set of ECM protein (Hynes, 2002). The repertoire of integrins present at the membrane dictates therefore the extent to which a cell will behave on a specific matrix and respond to its environment. Once engaged with the ECM, integrins cluster and recruit various signaling and adaptor proteins to form focal adhesion complexes (Geiger et al., 2001). These complexes activate intracellular downstream signaling pathways including NF-κB, PI3K, Src, or Ras-MAP kinases (Hynes, 2002; Legate et al., 2009). Such pathways regulate functions involved in motility, cytoskeleton organization, adhesion, proliferation, survival, and gene transcription. Integrins link ECM to the actin cytoskeleton through FAK/ILK/SFK/Rho proteins pathway providing the traction necessary for cell motility (Geiger et al., 2001). Integrins regulate the localization and the activity of urokinase-type plasminogen activator (uPA)/uPA receptor (Ghosh et al., 2000; Wei et al., 2007; Bass and Ellis, 2009) and matrix metalloproteinases (MMPs; Lamar et al., 2008; Morozevich et al., 2009) therefore controlling ECM remodeling and the invasive process.

Beside their mechanical functions and despite the lack of intrinsic kinase activity, integrins are true signaling molecules. Integrins regulate proliferation by controlling the expression of cyclin D1 which permits cells to enter the S-phase of the cell cycle (Fournier et al., 2008). Integrins relay survival or apoptotic signals depending on the surrounding environment. Integrin ligation promote survival through various mechanisms including increased anti-apoptotic proteins (bcl-2, FLIP; Aoudjit and Vuori, 2001b; Matter and Ruoslahti, 2001; Uhm et al., 1999), activation of PI3K–Akt (Aoudjit and Vuori, 2001a) or NF-κB pathway (Scatena and Giachelli, 2002; Courter et al., 2005). Unligated integrins were reported to promote apoptosis through the so-called integrin-mediated death (IMD) a mechanism dependent, or not, on caspases activation in anchorage-dependent cells (Stupack et al., 2001; Jan et al., 2004). However, tumor cells are often IMD-resistant and unligated integrins rather promote anchorage-independent growth, survival, and metastasis than apoptosis (Desgrosellier et al., 2009). Additionally, cross-talks occur between integrins, cytokines, and growth factor receptors. Optimal growth factor stimulation relies on integrin-mediated adhesion to an appropriate ECM protein. α_vβ_3/5 and α₅β₁ interact with growth factor receptors (VEGFR2, c-Met, FGFR1, PDGFR, EGFR, TIE-2, and IGF-1R) to promote full activation of each receptor and maximal signal transduction (increased MAPK and Akt activity) resulting in enhanced cell migration, proliferation, survival, and angiogenesis (Friedlander et al., 1995; Eliceiri, 2001; Alam et al., 2007; Soung et al., 2010). Integrins were also reported to bind directly growth factor (such as angiopoietins or VEGF) allowing the transduction of information in the absence of the receptor (Carlson et al., 2001; Hutchings et al., 2003). In short, integrins sense, interpret, and distribute information so that cancer cells adjust and respond to their microenvironment.

Clustering of transcriptomic data from high grade glioma predicted poor survival in subclasses of tumors overexpressing ECM components such as fibronectin which is the preferred ligand of α₅β₁ and α_vβ₃ integrins (Geiger et al., 2001; Freije et al., 2004; Bredel et al., 2005; Colin et al., 2006; Tso et al., 2006). Functional analysis revealed gliomagenesis and glioblastoma networks composed of genes that play a role in integrin signaling including fibronectin, α₃ and α₅ integrins (Bredel et al., 2005). Gingras et al. (1995) investigated glioblastoma for the expression of cell adhesion molecules including integrins that might distinguish tumor from normal adjacent brain tissue. Results showed that glioblastoma expressed α₂, α₃, α₅, α₆β₁, and α_vβ₃ integrins at significantly higher level than normal brain tissue suggesting that these integrins might play a role in the development or the progression of glioma (Gingras et al., 1995). β₈ and α₅β₁ integrins were commonly expressed in a perinecrotic or perivascular pattern in glioblastoma (Riemenschneider et al., 2005). Higher levels of α₅ and β₃ integrin mRNA were measured in glioblastoma as compared to normal brain or low grade astrocytoma (Kita et al., 2001). Average α_vβ₃ integrin expression in glioblastoma seemed to exceed those in low grade glioma at the protein level although mRNA levels of both subunits were not discriminative between glioblastoma and low grade glioma (Schnell et al., 2008). In another study, α_vβ₅ and α₅β₁ integrins were shown to be expressed at consistently higher levels than α_vβ₃ integrins in human glioma cell explants (Mattern et al., 2005). We and others showed recently that α₅β₁ integrin expression in biopsies from patient with glioma correlated with poor prognosis and tumor aggressiveness (Cosset et al., 2012; Holmes et al., 2012; Janouskova et al., 2012).

In glioma, integrins were often studied because of their crucial role in tumor cell invasion (Gritsenko et al., 2012). Both β₁ subunit-containing integrins (Paulus et al., 1996) and α_vβ₃ integrins control glioma cell invasion (D’Abaco and Kaye, 2007). α₂β_1, α₃β₁, and α₅β₁ integrins are overexpressed in multidrug-resistant glioma cells and are responsible for their increased adhesive and invasive capacities (Hikawa et al., 2000). The laminin-5 receptor α₃β₁ integrin is mainly expressed in areas of tumor cell invasion and support glioma cell migration and invasion (Tysnes et al., 1996; Mahesparan et al., 2003; Kawataki et al., 2007). α_vβ₃ integrins promote migration and adhesion in various glioma cells (Gladson and Cheresh, 1991; Friedlander et al., 1996) and inhibition of α_vβ₃ integrin with neutralizing antibody inhibited migration and invasion selectively in cell lines that contained a high level of integrin expression (Wild-Bode et al., 2001). Our laboratory extensively investigated α₅β₁ integrins in glioma. We showed that α₅β₁ integrins increased proliferation, clonogenic survival, adhesion, migration, and invasion of various glioma cell lines (Maglott et al., 2006; Bartik et al., 2008; Martin et al., 2009; Cosset et al., 2012). We also reported that the expression of α₅β₁ integrins in glioma is controlled by caveolin-1 (Martin et al., 2009; Cosset et al., 2012). Interestingly, it was shown recently that invasive recurrent glioblastoma, resistant to antiangiogenic therapy, overexpress α₅β₁ integrin and its ligand fibronectin (DeLay et al., 2012). α_vβ₃ integrin/ILK/RhoB pathway (Monferran et al., 2008) and β₁ integrin/AKT/p130Cas/paxillin (Cordes et al., 2006) controlled the radiosensitivity of glioma cells by regulating radiation-induced cell death. Recruitment of α_vβ_3/5 integrin in glioblastoma cells is induced by hypoxia. It follows the activation of FAK/RhoB/GSK3β pathway leading to HIF-1α induction and the transcription of proangiogenic factors (Skuli et al., 2009). Finally, surface expression of α₆β₁ integrin in U87MG cells enhanced cell spreading and attachment on laminin-111, increased proliferation, decreased apoptosis due to serum starvation and increased migration and invasion of U87MG cells both in vitro and in vivo (Delamarre et al., 2009).

Induction of angiogenesis is essential for a tumor to grow beyond 1–2 mm and glioblastoma exhibit prolific angiogenesis. Poorly expressed in resting endothelial cells, α₅β₁ and α_vβ_3/5 integrins are highly upregulated on endothelium cells during tumor angiogenesis (Bussolati et al., 2003; Avraamides et al., 2008; Desgrosellier and Cheresh, 2010) and rapidly accessible in tumor blood vessels (Magnussen et al., 2005). They stimulate endothelial cell proliferation, promote migration, and lumen formation (Mettouchi and Meneguzzi, 2006). Although α₅β₁ integrin is undoubtedly recognized as a proangiogenic factor, controversial results for αvβ3 integrin questioned its role (Hodivala-Dilke, 2008; Robinson and Hodivala-Dilke, 2011). Overexpression of α_vβ₃ integrin in glioma exerted growth-suppressive effects in vivo that are linked to vascular defects (Reynolds et al., 2002; Kanamori et al., 2004, 2006). However, antagonists to both α₅β₁ and α_vβ_3/5 integrins are able to inhibit tumor angiogenesis (Brooks et al., 1995; Friedlander et al., 1995; Kim et al., 2000).

Brain tumors also contain highly tumorigenic and therapeutically resistant pluripotent stem cells referred as glioma stem or initiating cells. The glioma stem cell hypothesis incorporates a model in which only a small subset of cells, the glioma stem cells, can initiate tumor. This hypothesis was confirmed very recently in vivo (Lathia et al., 2011). Elevated levels of α₆β₁ integrins were found in glioma stem cells and seem to be a reliable new marker to enrich for glioma stem cells (Lathia et al., 2010).

Integrins are implicated at various levels of glioma development and progression. Blocking their functions may affect both tumoral cells and endothelial cells and these characteristics made them attractive therapeutic targets for glioblastoma (Chamberlain et al., 2012; Goodman and Picard, 2012). Emphasis on α_vβ₃ integrins has been given recently as cilengitide, their prototypical small peptide antagonist, is currently evaluated in phase III clinical trials in glioblastoma (Tabatabai et al., 2010). Interestingly, the outcome in a phase II trial was particularly good in patients with a methylated MGMT gene promoter (Stupp et al., 2010). Emerging data showing the role of α₅β₁ integrin in glioblastoma give some hope for new therapeutic propositions in the near future.

As most mutations in p53 gene led to the accumulation of p53 in the nucleus, nuclear overexpression of p53 was usually considered as a marker of mutation. Several studies showed that the expression of p53 is correlated at 90% with its mutation (Figarella-Branger et al., 2011). Detection of p53 mutation by the yeast functional assay that measures quantitatively mutant p53 alleles and qualitatively the loss of p53 competence was also employed and compared to conventional techniques including DNA sequencing (Tada et al., 1997; Fulci et al., 2000). Overall results indicate that p53 mutations often occurred in low grade gliomas (WHO grade II astrocytoma; Bourne and Schiff, 2010) and thus is a frequent event in the pathological progression of secondary glioblastoma (WHO Grade IV; Gladson et al., 2010). Secondary glioblastoma arise from a preexisting grade II or III astrocytoma in contrast with primary glioblastoma that form de novo. Primary glioblastoma represent about 90% of glioblastoma. p53 gene mutations are present in about 30% of primary glioblastoma, and occur more frequently in secondary glioblastoma (65%; Ohgaki et al., 2004; Zheng et al., 2008). A recent integrated genomic analysis identified four relevant subclasses of glioblastoma (proneural, mesenchymal, neural, and classical glioblastoma). p53 mutation was observed in 54, 32, 21, and 0% of tumors from the proneural, mesenchymal, neural, and classical glioblastoma subtype, respectively (Verhaak et al., 2010). Interestingly, classical glioblastoma benefit from more aggressive therapy regimen than the others (Verhaak et al., 2010). In fact, the prognostic value of p53 status may be reconsidered according to these data.

No clear consensus has been reached about the prognostic value of p53 status despite numerous studies (Table 1). A clear picture remains difficult to draw due to the different techniques used to evaluate p53 (including immunostaining on tumor tissues, direct sequencing of p53 gene, and functional assays) and the complexity of patient cohort composition. Data illustrating an association of p53 with survival always point to a longer survival when p53 is mutated (Tada et al., 1998; Schiebe et al., 2000; Birner et al., 2002; Burton et al., 2002). However, the majority of studies do not validate p53 as an independent prognostic marker for glioblastoma (Kraus et al., 2001; Simmons et al., 2001; Shiraishi et al., 2002; Rich et al., 2005; Ruano et al., 2009; Weller et al., 2009; Levidou et al., 2010; Rossi et al., 2011). Overall it means that the prognostic impact of p53 aberrations is only marginal when considered in a global glioblastoma patient population. Reevaluation of this impact in clinically relevant glioblastoma subpopulations (see above) and association with specific molecular signatures will certainly be of interest in the future.

Recent studies begin to shed light onto the role of p53 in the regulation of neural stem cells (NSCs). NSCs are self-renewing cells in the central nervous system that can generate both neurons and glia. An elegant study showed that dual inactivation of p53 and PTEN in murine NSC promotes an undifferentiated state with high renewal potential and generates tumors with a high grade glioma phenotype (Zheng et al., 2008). Although the role of p53 in brain tumor stem cells has not been well established, data suggest that loss of differentiation and increase in neurosphere renewal may be linked to the disruption of the p53 pathway in glioma (Molchadsky et al., 2010; Mendrysa et al., 2011; Spike and Wahl, 2011). To achieve a permanent eradication of brain tumors, it is noteworthy that glioma-initiating stem cells have to be considered and in this way their p53 status and functions need to be further explored.

Despite expressing mainly a wild-type p53 and thus being expected to be sensitive to DNA-damaging agents, primary glioblastoma resist standard therapies including chemotherapy with TMZ. This intriguing observation is in debate and the role of p53 status in response to TMZ has been largely addressed in preclinical studies. Conflicting results have been obtained (Table 2) and show either an improved capability of TMZ to inhibit cell viability when p53wt is functionally inhibited (Hirose et al., 2001; Xu et al., 2001, 2005a,b; Dinca et al., 2008; Blough et al., 2011) or a sensitization of cells to drugs when p53wt is functional (Hermisson et al., 2006; Roos et al., 2007). The former studies suggested that glioma cells with an intact p53 gene are selectively impaired in the proapoptotic functions of p53wt while retaining the potential to mediate relevant DNA repair and cell cycle arrest. Treatment with TMZ induced a persistent cell cycle arrest and an increase in p21 (a cell cycle regulator) in functional p53-expressing cells which showed morphological and biochemical features of senescent cells (Hirose et al., 2001; Martinkova et al., 2010). In cells impaired for p53 function or with a mutant p53, TMZ induced a transient cell cycle arrest and cell death via apoptosis or mitotic catastrophe (Hirose et al., 2001; Martinkova et al., 2010) as well as attenuation of DNA repair (Xu et al., 2005b). When TMZ-triggered apoptosis was reported for both p53wt and p53mutant cells, pathways involved differed with activation of the FAS apoptotic pathway or the mitochondrial apoptotic pathway, respectively (Roos et al., 2007). Thus adverse effects of p53wt activities are increasingly recognized and may participate in chemoresistance of diverse cancers including glioma (Kim et al., 2009; Martinez-Rivera and Siddik, 2012). In one recent report, the effect of p53 status on response to TMZ was explored in glioma-initiating stem cells. It was shown that tumor stem cells are resistant to TMZ when p53 is mutated and sensitive to TMZ when intact (Blough et al., 2011). These data add a new level of complexity in the relationship between p53 status and TMZ sensitivity in glioma.