Gliomas are the most common primary malignant tumors in the brain. Glioblastoma multiforme (GBM), also known as World Health Organization grade IV glioma, is a disease with a high mortality rate worldwide, for which there is currently no effective therapy (1). The current standard treatments for GBM include maximal surgical resection, radiotherapy and temozolomide chemotherapy. However, GBM tends to recur despite therapy, with a recurrence rate as high as 90%. The median overall survival is 15-18 months according to population-based studies, and <10% of patients remain alive at 5 years post-diagnosis (2).

GBM is a heterogeneous tumor that is characterized by high resistance to therapy, which is promoted by GBM stem cells (GSCs) (1). Cancer stem cells constitute a small cell population that is highly involved in the malignant behavior of numerous types of cancer (3), and they are defined by their functionality, including maintenance of stemness, tumor formation, tumor relapse and resistance to therapy (4). GBM follows a cellular hierarchy model, with GSCs at the top and differentiated offspring cells at the bottom of the model (5,6).

There are significant similarities between neural stem/progenitor cells (NSCs) and GSCs, such as the expression of stem cell markers (CD133, SOX2, oligodendrocyte transcription factor 2 and nestin), and the ability to differentiate into oligodendrocytes, astrocytes and neurons. However, GSCs are characterized by the expression of multiple lineage markers in one differentiated cell (7,8). GSCs harbor genetic abnormalities, which contribute to tumor invasion (9), angiogenesis (10) and radio-resistance (11). The main characteristics of GSCs are their capacity for self-renewal and differentiation (3). In a xenograft assay, GSCs displayed greater tumorigenic capacity compared with non-stem tumor cells (6). GSCs play a pivotal role in the growth and therapeutic resistance of adult human GBM (7), suggesting that GSCs may lead to tumor recurrence and, eventually, death. Thus, exploring the signaling pathways regulating GSC self-renewal and the design of therapies targeting these signaling pathways are important research objectives. In recent years, numerous different signaling pathways and potential therapeutic targets for GBM have been identified, including TGF-β, Notch, Wnt, Hedgehog (HH) and STAT3, which also play important roles in normal stem cell development and differentiation. The focus of the present review was these five key self-renewal GSC signaling pathways and the corresponding differentiation therapies, with the aim of providing novel insight and promoting advances in the clinical therapy of GBM.

The TGF-β family includes polypeptides that regulate GSC maintenance and tumor differentiation (12,13). In addition, TGF-β signaling is involved in carcinogenesis and tumor development (12).

The TGF-β family comprises TGF-βs, bone morphogenetic proteins (BMPs) and other associated proteins (14). The TGF-β signaling cascade is a linear pathway from type II to type I receptor kinase to SMAD activation, resulting in the transcription of target genes in the cell nucleus. On the cell surface, TGF-β ligands bind to the transmembrane receptor serine/threonine kinase (type I and II) complex, and then type II receptor kinases [BMP receptor (BMPR)II, activin receptor (ActR)II, ActRIIB, TGF-β receptor (TβR)II and anti-Müllerian hormone receptor] and trans-phosphorylate type I receptors [anaplastic lymphoma kinase (ALK); ALK5/1/2 for TβR and ALK2/3/6 for BMPR]. The consequently activated type I receptors trigger phosphorylation of SMAD and receptor-regulated SMAD (R-SMAD) (TGF-β, R-SMAD2/3/1/5; BMP, R-SMAD1/5/8), which then form a complex with common mediator SMAD4 (Co-SMAD). The activated R-SMAD/Co-SMAD complex translocates to the nucleus to regulate gene transcription. The activation of R-SMAD is inhibited by SMAD6 or SMAD7 (14,15). It was previously demonstrated that TGF-β signaling regulates cell fate (16), and that blockade of this signaling pathway can inhibit proliferation of cancer cells and the GSC subpopulation (17). Thus, targeting the TGF-β pathway may be a meaningful treatment for GBM. The current clinical trials on drugs targeting the TGF-β signaling pathway are summarized in Table I.

Notch proteins (Notch 1-4) are transmembrane receptors that mediate cell-cell signaling. Notch signaling can amplify and consolidate molecular differences, eventually dictating cell growth, proliferation, survival and differentiation. Notch activity affects cell differentiation, proliferation and apoptotic programs (41). Both the receptors and ligands of the Notch family are cell surface type I transmembrane proteins. Notch ligands include delta (Dl) and Serrate. Upon ligand binding, Notch receptors undergo three proteolytic cleavages. The first cleavage, S1, generates fragments and forms a heterodimeric receptor, which is inserted in the cell membrane (42). S2 occurs after the heterodimers bind to the ligand (Dl-like-1, -3 and -4, and Jagged-1 and -2). S3 is mediated by the γ-secretase complex, leading to the release of Notch intracellular domain (NICD) into the nucleus (43).

The Notch signaling pathway, including NICD, hairy/enhancer-of-split (Hes)1 and Hes related family basic helix-loop-helix (bHLH) transcription factor with YRPW motif 1 (Hey1), regulates cell stemness and differentiation. Activation of the Notch receptor rapidly inhibits the death of NSCs (44). Inhibitors and activators targeting Notch receptors and ligands that exert antitumor effects have been developed. Notch stimulation results in poorly infiltrative but highly vascularized grafts, in contrast to the highly infiltrative and poorly vascularized characteristics of GBM stem cells. This indicates that the Notch pathway is crucial for regulating GSC fate (45).

During the early stages of embryogenesis, Notch signaling serves as a critical quality control pathway to prevent premature neurogenesis and maintain pools of progenitor cells in the developing central nervous system. In the perinatal stages, Notch signaling increases progenitor cell proliferation and drives astrocyte differentiation, thereby serving a critical function in human brain development. The Notch pathway is involved in maintaining adult neural stem cells bivalently by promoting self-renewal and repressing differentiation (46,47). The activity of the Notch signaling pathway plays an instrumental role in regulating self-renewal and determining cell fate of normal NSCs (48). The Notch pathway is active in NSCs during neurogenesis, gliogenesis and tumorigenesis (49,50). It has been demonstrated that the Notch target genes Hes1 and Hes5 are strongly associated with the regulation of neurogenesis and gliomagenesis in the brain (51).

The self-renewal capacity of GSCs relies on the activity of the Notch signaling pathway. The expression level of the Notch receptor gene and its downstream activation cascade of events are associated with the phenotypic plasticity and intratumor heterogeneity of GBM cells (49). A previous study that used computational modeling methods demonstrated that the stem cell renewal induced by the Notch pathway and the antagonistic effects exerted on the p53 pathway are highly involved in maintaining the regenerative properties of the NSCs (49,52). In agreement with this, previous in vitro and in vivo studies on glioma cell lines have indicated that CD133-positive GSCs are particularly sensitive to γ-secretase inhibitors or Notch1/2 knockdown compared with CD133-negative glioma cells (53). Blocking Notch signaling or recombination signal-binding protein for immunoglobulin κJ region (RBP-κJ), which is a major transcriptional effector of this pathway, reduced clonogenicity potential in tumor sphere assays and engraftment capacity in glioma xenograft models (48). Notch activity may contribute to intratumor heterogeneity by promoting stem cell behavior in poorly differentiated subpopulations of glioma cells. Notch signaling potentially regulates multiple steps of gliomagenesis, including tumor initiation, progression and recurrence.

However, the actual sequence of regulating events and the exact mechanisms through which Notch activity controls stemness and tumorigenicity remain to be elucidated. Since Notch can promote and maintain the stem cell characteristics of brain tumors, it may represent a promising target for developing more effective therapies against glioma. A phase I clinical trial investigating the use of γ-secretase/Notch inhibition in combination with temozolomide and radiotherapy in newly diagnosed GBM or anaplastic astrocytoma demonstrated that the addition of Notch inhibition to standard treatment was associated with certain benefits (Table II) (54), although it also clearly demonstrated that Notch inhibition, alone or combined with radiation and chemotherapy, may be insufficient for fully controlling tumor progression. However, those findings indicated that Notch may serve as a targeted biological tool that counteracts tumor stem cell-like behavior by preventing self-renewal and, possibly, angiogenesis (55).

The canonical Wnt/β-catenin pathway is a highly evolutionarily conserved signaling pathway that regulates pluripotency in stem cells (68). Frizzled proteins are receptors involved in Wnt signaling (69) that interact as co-receptors with Arrow, a low-density lipoprotein receptor-related protein that forms part of a receptor complex with Frizzled protein (70). When the Wnt ligand is activated by binding to the co-receptors, the AKT/glycogen synthase kinase (GSK)3β/adenomatous polyposis coli (APC) complex separates. The AKT/GSK3β/APC complex promotes the degradation of β-catenin, an intracellular signaling molecule. APC antagonizes the Wnt signaling pathway directly at the β-catenin effector level in several different ways: It acts as an adaptor between β-catenin and C-terminal binding protein, removes β-catenin, abrogates transcriptional transactivation, and inhibits the binding of the lymphoid enhancer-binding factor (LEF)/T-cell factor (TCF) proteins to β-catenin (71,72). Therefore, through the degradation of the AKT/GSK3β/APC complex, intracytoplasmic β-catenin becomes stable, and non-phosphorylated β-catenin accumulates in the cytoplasm and translocates to the nucleus to facilitate the transcription of target genes by interacting with the TCF and LEF transcription factors (71). The XTC-3 transcription factor mediates β-catenin-induced axis formation in Xenopus embryos. Functional interaction of β-catenin with the transcription factor LEF-1 leads to the nuclear localization of β-catenin (73-75). Wnt plays a key role in maintaining stemness in GBM cells (76,77). Thus, abnormal activation of the Wnt pathway may promote GSC self-renewal.

Classical HH signaling is required to maintain stem cell niches in the adult brain (95). HH has three gene homologs: Sonic HH (SHH), Desert HH and Indian HH. Upon inhibiting SHH signaling, the number of neural progenitors is reduced. Activation of the HH protein requires Rasp-dependent acylation (96). HH ligands initiate signaling pathways by binding to the transmembrane receptor protein patched homolog (PTCH). The HH-PTCH complex is internalized, and the inhibition of the receptor Smoothened (Smo) is abolished, thus allowing Smo activation, which induces the activation of the glioma-associated oncogene homolog (Gli) family. As a result, Gli translocates to the nucleus to regulate the transcription of target genes (97).

The STAT3 signaling pathway is involved in multiple biological processes, including cell proliferation, differentiation and self-renewal of GSCs. Cytokines and growth factors bind to their receptor, which, once dimerized, activates Janus kinase (JAK). JAK induces STATs phosphorylation, and activated STATs translocate into the nucleus to regulate target gene expression. Previous studies found that phosphorylated STAT3 interacts with the switch/sucrose non-fermentable complex in the nucleus (91,103). TRIM8, the expression of which is highly correlated with stem cell markers, is reported to activate STAT3 signaling to maintain the stemness and self-renewal of GSCs. TRIM8 activates STAT3 by suppressing the expression of the protein inhibitor of activated STAT3, and STAT3 activation can upregulate TRIM8, demonstrating that bidirectional TRIM8/STAT3 signaling is involved in the regulation of the stemness of GSCs (104).

Tetraspanin CD9, a regulator of cell adhesion, stabilizes the IL-6 receptor glycoprotein 130 (gp130) by preventing its ubiquitin-dependent lysosomal degradation, thus promoting bone marrow tyrosine kinase gene on chromosome X/STAT3 signaling in GSCs. Disrupting CD9 or gp130 can inhibit the self-renewal of GSCs and promote their differentiation (105). Currently, there are various ongoing clinical trials in USA investigating the targeting of STAT3 with the small molecule inhibitor WP1066 (Table III).

ID1 is highly expressed in GSCs and is involved in the TGF-β, Wnt and SHH signaling pathways. ID proteins are transcriptional regulators that are implicated in cell fate determination and differentiation of stem-like cells (106). Ubiquitination-specific proteases and cyclooxygenase-2-derived prostaglandin E2 have been reported to positively regulate the stability of ID1, and to promote GSC maintenance and treatment resistance (107,108). ID1 induces cell proliferation and promotes self-renewal through increasing cyclin E, the target molecule of cullin 3. Cullin 3 interacts with Gli2 and dishevelled segment polarity protein 2, and induces their degradation through ubiquitination. Loss of cullin 3 is the common signaling node in the Wnt and SHH signaling pathways through ID1 (109).

ID1 was previously found to inhibit BMP-mediated GSC differentiation through BMPRII and to maintain GSC traits (110). BMPs bind to a cognate high-affinity type II receptor (BMPRII) to phosphorylate the type I receptor (BMPRI). Activated BMPRI initiates downstream signaling by phosphorylating R-SMAD. ID1 could decrease BMPRII expression and the phosphorylation of its downstream signaling molecules SMAD1, SMAD5 and SMAD8 in cells (15,16). These results indicate that targeting ID1-driven intrinsic stemness signaling may be an effective therapeutic strategy for GBM.

GBM is a primary brain tumor with a high mortality rate, for which there is currently no effective therapy. Previous studies have found that GSCs promote the heterogeneity and treatment resistance of GBM. The main characteristics of GSCs are their capacity for self-renewal and differentiation. Therefore, elucidating the mechanism through which GSCs regulate the self-renewal response is meaningful in order to design therapeutic approaches targeting the self-renewal signaling pathways. The focus of the present review was five key self-renewal GSC signaling pathways, including TGF-β, Notch, Wnt, HH and STAT3, and the corresponding therapeutic targets, and the aim was to provide novel insight to enable advances in clinical therapy. Among these signaling pathways, gremlin 1, HMOX1, CXCL12/CXCR4, AEG1, CD163/CK2, ING5, CypA, HDAC6, CD9 and TRIM8 can increase stemness, promote self-renewal and prevent differentiation of GSCs. Therefore, their corresponding inhibitors may represent a novel type of therapeutic approach to glioma. On the other hand, Snail, FHL3, TRIM3 and CPEB1 may promote differentiation of GSCs. Thus, their corresponding agonists should be further investigated in this context. The aforementioned self-renewal pathways and corresponding differentiation-targeting treatments of GSCs are summarized in Fig. 1. Considering its complexity, the crosstalk between these pathways is not shown in Fig. 1. The current clinical trials targeting the TGF-β, Notch and STAT3 pathways are summarized in Tables I-III.

Based on the aforementioned findings, there are numerous potential treatments that are currently being explored. There are also emerging signaling pathways under investigation that may uncover potential treatment targets for GBM. In GSCs, lysine demethylase (KDM)1A (111), the transcription factors forkhead box G1 and transducin-like enhancer of split 1 (40,112), hypoxia-inducible factors (113), proliferating cell nuclear antigen (PCNA)-associated factor (PAF) (114), MEK partner-1 (MP1) (115), erythropoietin-producing hepatocellular receptors (116), progranulin (PGRN) (117) and DNA polymerase delta subunit 2 (POLD2) (118) are all overexpressed, and maintain GSC self-renewal capacity and stemness. In addition to the aforementioned classical pathways, the molecular mechanisms through which these factors maintain stemness require deeper and more comprehensive investigation. For example, PAF promotes the maintenance of self-renewal ability and stemness by interacting with PCNA, and regulates PCNA-associated DNA translesion synthesis (114), while MP1 contributes to GSC stemness by driving ERK activity (115).

Other factors play a unique role in the damage and repair of DNA, such as POLD2 and PGRN (117,118). Previous studies have demonstrated that PGRN promotes DNA repair through activator protein 1 transcription factor, cFos and JunB (117). In terms of their relevance to treatment, the knockdown of these molecules can reduce GSC stemness and induce their differentiation. Based on the identification of these factors that maintain stemness, corresponding inhibitors may be developed to target GSCs. For example, a series of inhibitors have already been developed and evaluated. Two novel KDM1A-specific inhibitors (NCL-1 and NCD-38) were found to significantly reduce GSCs-driven tumor progression by inducing the activation of the unfolded protein response pathway (111). GLPG1790, a small-molecule ephrin receptor inhibitor, completely blocks ephrin type-A receptor 2 signaling and exerts antitumor effects (116). Similarly, GSC gap junctions also have pro-tumorigenic effects depending on connexin expression (119). However, to the best of our knowledge, the detailed mechanisms remain elusive and further research is needed in this field.

In addition to the aforementioned factors that maintain stemness, other factors promote differentiation, and regulating their activity may be of value in the context of differentiation therapy. For example, MAPK phosphatase 1 (MKP1), a dual-specificity phosphatase, acts as a negative inhibitor of JNK, ERK1/2 and p38 MAPK. High levels of MKP1 expression impair self-renewal and induce differentiation in GSCs (120). The let-7 miRNA family has also been shown to induce GSC differentiation. The mechanism is as follows: Its recognition elements may be bound by insulin-like growth factor 2 mRNA-binding protein 2, which prevents let-7 target gene silencing and impairs the maintenance of GSC stemness (121).

In summary, inhibitors of the factors found to maintain stemness may be developed in the future to provide possible differentiation therapies. For the factors that can promote differentiation, increasing their expression levels is an important method for targeting GSCs. It is expected that more clinically feasible differentiation treatments will be developed in the future in order to improve GBM treatment efficacy and prognosis.