Glioma: molecular signature and crossroads with tumor microenvironment

EGFR is a receptor tyrosine kinase located as a transmembrane protein. It is activated extracellular by ligand binding and intracellular via autophosphorylation of the kinase domain via downstream signaling pathways [26], thereby stimulating cell proliferation and survival. Tyrosine hydroxylase is induced by EGFR as shown in clonal pheochromocytoma cell line [54]; its overexpression leads to decreased interferon (IFN)-γ and tumor necrosis factor (TNF) levels as shown for lymphocytes [23], which could be a marker for poor prognosis [55], and activates the xc(-) system [56]. Xc(-) is an amino acid antiporter that regulates the exchange of extracellular l-cysteine and intracellular l-glutamate and regulates oxidative stress in various cell types [57, 58] (Fig. 2). An increased EGFR expression, in turn, leads to an increased glutamate concentration in the TME (Fig. 2), which promotes glioma cells migration via phosphorylation of the COOH terminal (carboxyl-terminus) of GluN2B (glutamate (NMDA) receptor subunit epsilon-2), resulting in increased glutamate-NMDAR-signaling [12]. Conversely, in the presence of the NMDAR-inhibitor MK-801 and sulfasalazine (an inhibitor of xc-), glioma cell migration was not observed [12]. In turn, the active mutation EGFRvIII that is transmitted into the TME from glioma cell EVs [59] acts pro-tumorigenic and supports glioma cell survival, angiogenesis, invasion, and stemness [24]. Notwithstanding, EGFRvIII is associated with increased overall survival in patients, probably caused by inflammatory triggering properties of EGFRvIII [26]. Moreover, it could be recently shown that GBM cell lines expressing endogenous EGFRvIII increased sensitivity to temozolomide in mice, which might be linked to increased DNA (deoxyribonucleic acid) mismatch repair proteins [27]. In contrast, however, temozolomide can presumably also exert selection pressure on the molecular properties of the glioma cells. For example, mutations of the mechanistic target of rapamycin (mTOR) or platelet-derived growth factor receptor (PDGFR) occur [60], leading to tumor recurrences through a “molecular resistance.” Cooperation between EGFR and EGFRvIII induces macrophage infiltration via elevated chemokine CCL2 (CC-chemokine-ligand-2) [24]. Kirsten rat sarcoma virus gene (KRAS) is involved, a monomeric G-protein and actor in signal transduction pathways for the differentiation and growth of cells. CCL2 is called a “tumor-derived chemotactic factor.” It recruits several classes of immune cells like monocytes, dendritic cells, memory T cells, and natural killer cells (Fig. 2), whereby pro-inflammatory mechanisms are modulated and neo-angiogenesis is increased [61]. Moreover, glioma expressing EGFRvIII secreting IL-6 is associated with high pro-angiogenic IL-8 [62, 63]. IL-6 activates the NF-κB (nuclear factor “kappa-light-chain-enhancer” of activated B-cells) pathway, which makes the cells less sensitive to attenuation of EGFR tyrosine kinase inhibitors [64] and induces the upregulation of programmed death-ligand (PD-L1) on peripheral myeloid cells, which promotes tumor growth in orthotopic mice models [65]. The expression of PD-L1 in high-grade glioma induces PD-L1 receptor expression in microglia, which then causes, immunosuppression in the sense of a negative T-cell response [66]. Additionally, via NF-κB signaling, antioxidant genes are regulated in cooperation with transcriptional coactivators in inflammation [67] (Fig. 2). Thus, via EGFR (vIII), glioma cell migration and survival are supported, as well as a change of the immunological state of TME in terms of a weakened proper immune response. Angiogenesis and stemness are promoted via interleukin mediation. EGFR directly influences a variety of aspects that mediate the glioma process and is a central mediator in glioma cell-TME interaction.

The mutation status (-mut) of the genes encoding isocitrate dehydrogenase (IDH) 1/2 is of central importance for the differentiation of low-grade to higher-grade gliomas, as low-grade gliomas present higher IDH-mut rates than high-grade gliomas [68]. IDH1mut is associated with a favorable response to chemotherapy and radiation [69], and 70–90% of low-grade-glioma and secondary glioblastoma present mutations in IDH1 [70]. Non-mut-IDH converts isocitrate to α-ketoglutarate. α-Ketoglutarate is used to catalyze 2-hydroxyglutarate, an oncometabolite, and is common in IDH-mut [71]. The immune system of the TME appears to be different between IDH-mut and IDH-wt-gliomas. More tumor-infiltrating lymphocytes are present in the latter, and PD-L1 is expressed higher [28]. Interestingly, the reduced immune situation seems to be associated with an IDH-1 mutation since the expression of the genes for IFN-γ and CD8⁺ is reduced. Both genes are essential for an antitumor T-cell response [29]. This is astonishing since a weakened immune system results in a molecular subtype that goes hand in hand with a better prognosis. Derived from this, this would mean that IDH-1 wild-type tumors, for instance, most higher-grade gliomas, would be more accessible for immunotherapy. As mentioned, distinguishing low-grade and higher-grade gliomas 1p/19q-status (loss of heterozygosity (LOH)) and IDH-mut is crucial as a co-deletion of 1p/19q is typical for oligondendroglioma and essential for the differentiation against astrocytoma. Furthermore, the co-deletion indicates a better response to chemotherapy. Co-deletion of 1p/19q is mainly associated with IDH-mut, but IDH-mut does not always belong to a 1p/19q deletion [72]. It also appears that the 1p/19q status is related to the expression of integrins. These transmembrane receptors not only play a role in the adhesion of cells with the extracellular matrix but are also involved in processes such as the regulation of induced cell death, intracellular transport mechanisms, or migration in conjunction with growth factors [73]. For over 20 years, it is known that α6β4 integrin has significantly increased in human astrocytomas [74]. Recently, it was reported that in the presence of 1p/19q deletion and IDH1-mut, the expression of integrin beta-1 (ITGB1) was reduced compared to IDH1-mut, but a knockdown of α6β4 integrin in 1p/19q intact glioma cells (in vitro) and patient-derived glioma cells led to a decreased glioma development [75] (Fig. 2). IDH mutation is one of the most important genetic modification in glioma, as it supports early spread of glioma cells and, therefore, is an important early marker and therapeutic target. However, the effect on the immune-tumor microenvironment is also observed by inhibiting the T-cell response and the association with PDL-1.

ATRX (ATP-dependent helicase) and TERT (telomerase reverse transcriptase) mutations are usually present together and are essential for further differentiation of gliomas. The mutations are mostly linked to IDH-mut and an increased EGFR presence [76]. A mutation in the TERT promoter is associated with an increased TERT expression. Conversely, (inactivating) mutations for ATRX lead to a reduction of ATRX in glioma cell nuclei, which means that gene regulation via chromatin remodeling is no longer regulated [30], thereby promoting stabilization of the glioma cell. ATRX is entangled in the misregistration of immune signals in tumor cells [31]. For cell lines, it is showed that deletion of ATRX presumably promotes the expression of (type I) interferon. The authors also report better survival time in a xenograft and a syngeneic murine glioma model (ATRX KO cell line) and suspect an interaction between ATRX mutation and immune signaling in glioma cells [31] (Fig. 2). Thus, ATRX is a central player of glioma cell genesis. Depending on the mutation status, there is an influence on the stabilization of tumor cells and the presence of interferon (type I). As an important marker, ATRX is concluded in the WHO Classification 2016 (oligodendroglial vs. astrocytic) and seems to be involved in the mediation of TME immunological parameters.

Clinical therapies have been established with vemurafenib, a selective inhibitor of BRAF (v-raf murine sarcoma viral oncogene homolog B) and dabrafenib in treating malignant melanoma, as well as non-small cell bronchial and thyroid carcinoma. BRAF mutations are detectable in various tumors and are encoded on chromosome 7 (7q34). The protein is primarily responsible for controlling cell growth signals, including regulating the MAP kinase signaling pathway. Acquired mutations are primarily involved in the participation in tumor growth [77]. The common BRAF V600E mutation and the KIAA1549-BRAF fusion are present in gliomas [32]. The detection of a fusion gene in combination with an IDH-mut analysis is essential for clinical diagnostics, especially for the differentiation between pilocytic astrocytomas (WHO I) and diffuse astrocytomas (WHO II) [77]. A BRAV V600E mutation is primarily found in pleomorphic xanthoastrocytomas (approximately 60–70%) and gangliogliomas (about 20%) in diffuse, higher-grade gliomas [78, 79]. However, for pilocytic astrocytomas, it is reported that BRAF activation promotes pro-cancerogenic senescence via a p16 (INK4a) pathway and causes aberrant activation of the protein kinase pathway [32]. The KIAA1549-BRAF fusion creates a pro-cancerogenic TME via the CCL2/CCR2 axis, mediated by NF-κB and microglia recruitment [33] (Fig. 2).

NF1 (neurofibromatosis type 1) is a broad predisposition syndrome for various tumors, also for gliomas and most frequent tumor suppressor syndrome [80, 81], with an incidence of 1/3000 births [81] and a regulator of the RAS signaling pathways, essential for regulating cell differentiation and growth. A loss of neurofibromin expression promotes increased cell growth—as in tumors associated with NF1—via increased RAS-activation [80]. For glioblastoma cells, it has been shown that NF1 incompetence led to a decreased cancer cell homogeneity, enhanced NF1 expression led to diminished microglia activity and vice versa, and NF1 deactivation results in increased macrophage activation [34] (Fig. 2). KIAA1549-BRAF fusion is a central player in achieving cell survival and cell proliferation and is, as described, an important clinical marker for the delimitation of pilocytic astrocytoma. The influence on the CCL2/CCR2 axis described here is important for the interaction with the TME since microglia recruitment occurs and pro-cancerogenic processes are initiated. It seems particularly interesting to decipher the processes that TME mediates on the microglia to pursue new therapeutic approaches and the mechanisms in the glial cells themselves that lead to glioma-like properties. The understanding of the influence of NF1 on glioma properties as a whole and concerning the tumor microenvironment is still minimal. The effect of NF1 to induce decreased microglial activity appears to be an interesting starting point for further research. It can be assumed that immunological processes, mediated by TME, occur here since NF1 deactivation is associated with increased macrophage activation. Immune cell analyses in animal models and in vitro can be a corresponding starting point for researching these processes in more detail.

The gene PTEN encodes the protein phosphatase and tensin homolog. Dysregulation of the gene is associated with different types of cancers [82]. A PTEN-deletion occurs in 30–40% of high-grade gliomas, but the impact on overall patient survival is discussed [83]. However, PTEN plays a vital role in preventing immunosuppressive TME and tumor cell evasion [35]. For glioma cells with PTEN mutation or deletion, an immunosuppressive TME evolved: the anti-cancerogenic immunity decreased, and resistance against T-cell lyses developed, together with a PDL-1 enhancement [35, 36]. In human glioblastoma cells, increased T-cell apoptosis could be detected in the presence of PTEN-deficient glioblastoma cells. Along with this, the absence of PTEN led via the PI3K-Akt-mTOR pathway to an immune resistance [35, 37]. For melanoma, it could be shown that PTEN-deficient cells encourage a resistance to immune infiltration by increased levels of cytokines, such as VEGFR and CCL2, resulting in an immunosuppressive tumor microenvironment [84, 85]. In glioma cells, PTEN deficiency fosters activation of yes-associated protein 1 (YAP1), which modulates the protein-lysine 6-oxidase (LOX) [38]. As a result, the conversion of molecules into activating aldehydes is induced, which are present in the extracellular matrix, promoting cross-linking of proteins such as elastin and collagen [39]. Tumor-infiltrating macrophages that secrete phosphoprotein 1 (SSP1) were increased via LOX activation, supporting glioma cell growth and angiogenesis. For glioblastoma cells that did not express PTEN, LOX was noticeably reduced and macrophage infiltration and tumor progression took place [38] (Fig. 2). Loss of the tumor progressor is of importance to therapeutic resistance. Still, the imparting of immunological characteristics of TME also seems to support glioma progress, such as the shift to a more immunosuppressive TME or immune resistance in the case of PTEN mutation and deficiency reveals. In addition to immune mediation, functional PTEN also appears to inhibit angiogenesis and the establishment of cross-link proteins and is, therefore, a possible key element of glioma cell stabilization.

DNA repair enzymes and proteins such as MGMT, ERCC (DNA excision repair protein), and APNG (alkylpurine-DNA-N-glycosylase) are also likely to be determined by extracellular mechanisms. If MGMT is hypermethylated, alkylating chemotherapy response and more prolonged survival are more likely [41]. MGMT, APNG, and ERCC1 mRNA were increased in microvesicles derived from glioma cells [86]. Understanding these mechanisms seems more important because the DNA repair enzyme MGMT inhibits the efficacy of standard clinical drugs such as temozolomide. A hypoxic TME promotes the expression of MGMT [42] (Fig. 3), and MGMT gene silencing in patients prolonged survival time under treatment temozolomide [87]. MGMT expression also depends on cellular-cellular signaling and was reduced in case of decreased Wnt signaling [43]. This downstream signal transduction pathway works in a canonical/non-canonical matter and is crucial in various diseases [88] (Fig. 3). MGMT status is crucial in the clinical evaluation of glioblastomas, as methylation status is directly associated with the response to alkylating therapies, especially in the elderly. However, a hypoxic tumor microenvironment appears to establish the therapeutically inferior MGMT status. The data on the influence of MGMT on immunological processes is still rudimentary, at least, as we have not been able to determine any valid data on this, but there are already hints that MGMT methylation and the immune response interact. Deciphering these processes is of particular clinical importance since temozolomide has immunosuppressive properties.

p53, a tumor suppressor, induces the arrest of the cell cycle. Thus, a mutation of the gene that encodes the tumor suppressor protein p53 limits the prognosis in cancer [89, 90] and is present in many tumor entities and approximately 30% in glioblastomas [91]. Dysfunction of the tumor suppressor promotes cell invasion and migration of glioma cells and has a protective effect against apoptosis [92]. However, the formation of the senescence-associated secretory phenotype (SASP) is described to be p53 dependent [93] (Fig. 3), and a loss of p53 could lead to pro-cancerogenic activities of SASP [45]. Interestingly, SASP is more present in high-grade gliomas—as SASP-related genes are in these grades more expressed, and an association with IDH-wt astrocytoma is assumed [72]. SASP is involved in transforming fibroblasts into active inflammatory cells, which support tumor growth [45]. Interestingly, SASP in tumor stroma can have an immunosuppressive effect on the tumor microenvironment [46], but so far, this has only been shown for liver cells [89, 90]. On the other hand, SASP seems to support tumor progression, as it can promote epithelial-mesenchymal transition [46]. Thus, the role of SASP in glioma seems not to be finally elucidated, but it suggests that the effect of SASP is related to p53 activity [94].

Further studies show that so-called gain of function (GOF) influences mutations, which, in turn, causes inflammatory processes in the glioma TME [91] and indicates that misregulation of p53 supports inflammatory processes and also tumor-immune evasion [47, 48] (Fig. 3). Glycoproteins of the extracellular matrix (ECM) also appear to interact with p53. In TME with tumor-infiltrating leukocytes, for example, it could be shown that the pharmacological activation of p53 led to an increased anti-cancerogenic immune system and tumor regression, in addition to immune activation through the induction of immune-associated cell death [48]. For astrocytes with a p53 + / − , fibronectin and laminin are more concentrated in the ECM than in the ECM of p53 + / + astrocytes. Astrocytes with a p53 + / − were also less susceptible to apoptosis in vitro [95]. This could be due to a haploinsufficient phenotype of p53 + / − and deduce that higher-grade glioma cells create a dysfunctional TME by the reduced expression of p53 [95]. In other studies, it could be shown that the presence of laminin in the ECM in principle enables better anchoring of the cells in their environment and counteracts cell migration and invasion [96]. However, the interrelationships between glioma and TME are clearly more complex; previous studies showed that both laminin and fibronectin were involved in tumor cell migration in vitro [97]. There is currently no detailed knowledge of the interrelations between ECM proteins and the p53 status in gliomas. With mutations of the p53 gene, many interactions with compartments of the TME seem to enter, which are pro-cancerogenic but also offer approaches for new therapies (Fig. 3). p53 is certainly one of the most important known factors in eukaryotic cells to prevent their degeneration into tumor cells. Conversely, dysregulation has correspondingly many effects on cell metabolism, apoptosis, or the stability of the genome. Interactions with the TME, as we have shown in this paragraph, are also versatile. Cell invasion and migration of glioma cells are promoted in the case of altered p53, as well immunosuppression. As in many other tumors, p53 seems to be a promising starting point for glioma therapies to enable the preservation of an anti-cancerogenic TME.

It has been known for many years that the malfunction of the tumor suppressor protein RB1 (RB transcriptional corepressor 1) is involved in many pro-cancerogenic processes [98], and astrocytomas expression of mutated RB1 is associated with proliferation of tumor cells and is correlated to decreased survival [50]. In non-cancer cells, RB1 is dephosphorylated by the cyclin-dependent kinase 4/6 (CDK4/6) pathway and can bind the transcription factor E2F, which prevents cell division. CDK4/6 dysfunction is common in high-grade glioma [99] and promotes phosphorylation of RB1. As a result, E2F binding fails and glioma cells’ division occurs [49]. Studies using fluorescence in situ hybridization and immunohistochemistry showed a complete loss of RB1 in about 10% of glioblastoma samples examined [50]. A mutation of the 13q14 gene encoding RB1 exists in approximately 30% of all higher-grade astrocytomas [100]. p21, an inhibitor of a cyclin-dependent kinase, supports a complex between cyclin D1 and CDK4 [51]. p53 controls transcription of p21 [101] is increased by DNA damage or cell stress (for example, through a hypoxic TME) and supports physiological cell cycle arrest in increased expression levels [102]. Cyclin D1, a regulatory subunit of CDK4/6 in turn, is increasingly expressed in glioma cells [103] (Fig. 3). Furthermore, CD1 and CDK4 are critical for both the glioma cell and the TME as the lack of CDK4 prevents glioma cell development, and the absence of CD1 alone had an opposite effect [51]. Interestingly, a combined loss of CD4 and D1 led to reduced microglial activation in the TME. In many cancer entities, the CDK4/RB1 pathway is disturbed and leads to increased cell proliferation. Preclinical therapeutic approaches with inhibitors foster cell cycle arrest of the tumor cells. However, a hypoxic tumor cell environment will affect p53 and p21 as described above. The latter, in turn, influences the proper function of CDK4/6 which, together with cyclin D1, co-determine the phosphorylation status of RB1. As far as we could determine, no direct interaction between (immune-) TME and CDK4/6 is known, but at least via p53/p21 CDK4/6 seems to be indirectly influenced by the conditions of the tumor microenvironment. Consequently, CDK4/6 inhibitors could be of particular interest in glioma patients with a disturbed p53/p21 axis.

The invasive growth of glioblastomas is significantly influenced by micro-anatomical areas with unique molecular properties and is crucial for stem cell differentiation, glioma cell invasion, and immune evasion. As early as 1938, the German neuropathologist, Hans Joachim Scherer, characterized the unique growth properties of gliomas concerning the white brain matter and the meninges. However, Scherer especially described perivascular tumor progress outside the Robin-Virchow areas [116]. Three major micro-anatomical niches of glioma TME can be divided, which play an essential role in the metabolic transformation of glioma cells and stem cell differentiation: the invasive, hypoxic, and perivascular niche [52] (Fig. 1). Multiple players are involved in glioma “behavior” as peripheral blood cells such as monocytes differentiate into tumor-associated macrophages and accumulate in the glioma as part of an inflammatory reaction. Furthermore, other immunological players such as inflammatory cytokines also influence glioma growth. As the tumor progresses, the glioma TME changes to an immunosuppressive environment, making immunotherapeutic approaches more difficult.

In all three niches, GSCs are involved in tumor development and stabilization, as well as resistance to therapies [2]. GSC’s formation and characteristics like self-replication, specialization, proliferation, and tumor induction might depend heavily on its environment. There may be very heterogeneous GSC variants in the TME, even those with glioma cell properties. In the perivascular and invasive niche, GSCs are in close contact with the excessive vascular structures and communicate with normal and aberrant blood vessels [52]. Moreover, GSCs and tumor cells can communicate with each other, thereby stabilizing themselves and supporting processes such as immune evasion that are an obstacle to tumor cell control, whereas the niches themselves appear to be dependent on the tumor entity [52]. The GSC regulation is endogenous and exogenous. In addition to metabolic regulation and genetic determinants, factors of the niches but also parameters of the TME such as the immune system are essential [2]. There are markers for many cancer stem cells, but they are variable. For GSC, the cell surface protein CD133 could be identified and was mainly represented on GSC, which showed a high proliferation [117]. However, the expression of CD133 is from the phase of the cell cycle, so that active GSCs can also display a low surface density of CD133 [2]. Other GSC markers such as integrin CD15/SSEA-1 and CD44α6 are not always reliable indicators of the actual characteristics of GSCs [2]. GSCs play a vital role in all niches for gliomas’ invasive growth, as shown in Fig. 1.

The perivascular niche with neuronal progenitor cells is characterized by increased angiogenesis and signaling protein vascular endothelial growth factor (VEGF). VEGF is supported by a mutation of the PTEN gene, which, together with EGFR, can induce increased VEGF mRNA levels [40]. However, VEGF is crucial for the enormous vascularity in glioma. It has been shown that brain tumor stem cells with the high expression level of CD133 habit also high VEGF level [118]. GSCs are also critical for tumor differentiation, which decisively influences glioma genesis due to their high differentiation potential. Those who possess the transmembrane protein CD133 contribute to angiogenesis since VEGF is associatively increased [52, 118, 119]. VEGF contributes to the blood–brain barrier breakdown by destroying tight junction proteins [120]. As a result, the brain’s privileged immune situation is lost, and peripheral immunomodulators can enter the central nervous system and the tumor more easily. The pro-inflammatory cytokine and an anti-inflammatory myokine IL-6 are associated with tissue destruction and accompanying inflammatory processes in the sense of an acute-phase reaction induced by TNF-α [121]. The cytokine aid the invasive behavior of high-grade-gliomas via PD-L1 induction via STAT-3 [65, 122], resulting in a negative T cell response. IL-8 is also a marker of pro-cancerogenic processes, supports the invasive properties of glioblastomas, and seems to mediate glioma cells’ plasticity. IL-8 is also attributed to shield glioma cells against a hypoxic environment and protects them from therapeutic stress [123]. A misdirected, IL-6-dependent NF-κB activation is associated with increased IL-8 levels [124].

Metabolic changes in the niche support the entire process of tumor invasion, survival, and growth. A hypoxic constitution of TME is a vast inductor of multiple pro-carcinogen pathway activities. Stem cells, also known as tumor-initiating cells, seem to adapt remarkably well to the hypoxic, lactate-rich environment, promoting tumor growth [25]. A central component of the regulation associated with HIF 1-alpha and HIF-2 alpha in the hypoxic niche represents angiogenesis and resistance to the acidotic environment, among other things, through the upregulation of VEGF and IL-8 and also supports stem cell presence as well as CD133, which is also increased in cells in hypoxic environments [52, 53]. Crucial pathways such as WNT are also upregulated in cells in a hypoxic environment and are associated with the unmethylated MGMT type, chemotherapeutic resistance, and reduced induction of cell cycle arrest, as described above. The hypoxic environment also leads to increased EGFR activity, which induces mTOR HIF [25]. Therefore, it could be assumed that mTOR-inhibiting chemotherapeutic agents can have a good advantage, especially on glioma cells in a hypoxic environment. Additionally, immunological processes are triggered by hypoxic conditions. The HIF expression induces activation of tyrosine hydroxylase [125], and a reduction of IFN-y and TNF can occur, which has been shown for lymphocytes [23]. A reduction in IFN-y has a pro-cancerogenic effect since it promotes apoptosis, restrains cell proliferation (shown in lung carcinoma), and has an antiproliferative influence on glioblastoma growth in cell lines [126, 127]. In turn, prolyl hydroxylase domain enzymes (PHD), which α-KG and Fe2⁺ require, are essential for the degradation and hydroxylation of HIF. In IDH-mut cells, there are decreased levels of α-ketoglutaric acid and Fe2⁺, which would support HIF [68]. In hypoxic TME areas, a glycolytic metabolism is supported, resulting in increased lactate and H⁺ ion levels. Both acidosis and hypoxia can affect the essential cellular organelle mitochondria by altering glycolysis [22, 128]. Thus, investigation of the exact mechanisms of glioma cell—TME interactions under hypoxic metabolic circuit—might increase the chances of developing new effective therapies.

An essential feature of glioblastomas is their highly invasive growth. Glioblastomas do not metastasize through lymphatic systems or vascularly. The spreading occurs along the vessels’ course; here, glioma cells displace the vascular astrocytes regardless of the vessel size [129]. GSCs stimulated by ligands such as notch, angiopeptin, or endocrine are released from vascular endothelial cells [60]. This is supported by the CXC chemokine receptor type 4 (CD184), encoded by the CXCR4 and the stromal cell-derived factor 1 (CXCL12). The CXCL12/CXCR4 axis drives cell proliferation, angiogenesis, and glioma invasion. Glioma stem cells are associated with endothelial cells using the CXL12/CXCR4 axis. Mediated by TGF-ß, the transformation to pericytes then takes place. Conversely, eradicating these pericytes inhibits the glioma process along the vessels by reducing neovascularization [130]. A current study showed that the forkhead box protein M1, a transcription factor involved in physiological cell division and cell cycle regulation [131], is increasingly expressed via the CXCL12/CXCR4 axis, thereby increasing resistance to temozolomide [132].