The tumor microenvironment/stroma is closely related to tumorigenesis, metastasis and invasion (55,56). The tumor microenvironment/stroma is mainly composed of endothelial cells, adipocytes, immune cells and cancer-associated fibroblasts (CAFs) (57). CAFs are able to secrete soluble factors to stimulate cancer cells, thereby triggering tumor metastasis and chemotherapy resistance (58–60). Recombinant human Sonic HH N-terminal peptide (rhSHH) enhances HH signaling, accompanied by increased mRNA and protein levels of matrix metalloproteinase-2 (MMP2) and MMP9. Furthermore, the protein expression of GLI1 was positively associated with the protein expression of MMP-2 and −9, which promoted the adhesion and invasion of GBM cells (60). It has been reported that gap junctions have a role in tumor growth and progression. Torrisi et al (61) modulated SHH signaling and connexin 43 (CX43)-based intercellular communication in an in vitro model. Modulation of SMO with the use of a known agonist (i.e., taxamine) and a known antagonist (i.e., cyclodopamine) affected CX43 expression levels and thus affected related functions. In addition, SMO activation also promoted cell proliferation and migration. Of note, inhibition of the CX43 channel prevented the SMO-induced effects (61).

Therefore, further exploration of the mechanisms of the HH signaling pathway in the tumor microenvironment may lead to better targeting of this pathway to fight cancer.

In the development of GBM disease, PC serve as cell antennae to transmit and regulate a variety of signaling pathways and SHH is one of the most important pathways. SHH levels are significantly increased in GBM cells compared with normal brain tissue and SHH overexpression induced neuroectodermal angiogenesis during mouse embryonic development (62–64). A study has indicated that Fms-related tyrosine kinase 1 (FLT1) is significantly increased in GBM cells and overexpression of FLT1 increased the expression of SHH in cells (64). Knockdown of SHH reduced the migration and invasion mediated by FLT1 overexpression, while overexpression of SHH restored the migration and invasive ability of FLT1 knockdown (64). FLT1 is a tyrosine kinase receptor that binds VEGF-A with several times the affinity of other kinases inserted into domain receptors and has been reported to promote tumor growth and metastasis (65). VEGF-A is one of the key factors promoting tumor angiogenesis and activation of the VEGF-A pathway requires the binding of VEGF-A to its receptor FLT1 to generate downstream signals to stimulate the proliferation and development of tumor cells and provide tumor blood vessels for the growth and metastasis of GBM (64,66). In addition, brain tumorigenic-initiating cells produce DHH ligands to realize the paracrine DHH/PTCH2 signaling cascade, transmit high permeability and angiogenesis, and also promote GBM growth (6). Chen et al (67) reported that C-terminal binding protein 2 (CtBP2) expression was increased and zinc finger and BTB domain containing 18 (ZBTB18) expression was decreased in GBM tissues, and the two were negatively correlated. CtBP2 short hairpin (sh)RNA interacts with ZBTB18 to block cells in G0/G1 phase, inhibit the SHH-Gli1 pathway and reduce the tumor volume (67). However, whether this effect is exerted by increasing SHH gene expression has remained to be elucidated. Therefore, targeted FLT1 or CtBP2 therapy may be a promising direction to develop anti-metastasis agents.

GBM develops through a complex interlocking signaling pathway. RhSHH enhances the HH signaling pathway, which increases the production of MMP-2 and −9 through the PI3K/AKT pathway, thereby regulating migration and invasion of basal membrane cells and promoting GBM cell adhesion, invasion and migration (60). By contrast, triggering the vasoactive intestinal peptide receptor system is triggered to reduce GBM cell migration and invasion through PKA-dependent PI3K/AKT and SHH/GLI1 pathway blocking (68). Similarly, Henao-Restrepo et al (69) reported that PI3K/AKT/mammalian target of rapamycin complex 1 (mTORC1) and SHH/GLI signaling pathway proteins were expressed differently in human gliomas with different tumor types and grades, suggesting that the activation of these signaling networks is related to the occurrence and development of high-grade gliomas.

Multiple studies have indicated that the SHH signaling pathway promotes the plasticity of cancer cells by regulating the adhesion between cells and the extracellular matrix, thus increasing the motility and aggressiveness of cells, leading to poor prognosis of patients (60,65,67,68). Statistical analysis of the The Cancer Genome Atlas (TCGA) dataset (TCGA-Glioblastoma June 2016) suggested that SHH upregulation was associated with decreased overall survival (64).

Hedgehog-interacting protein (HHIP), which is located on chromosome 4q31.21-31.3, is defined as an antagonist of SHH, IHH and DHH. Chang et al (9) were the first to demonstrate that the expression of HHIP determined by immunohistochemistry is an independent prognostic marker of favorable outcomes in patients with GBM.

Although GLI1 was originally identified as the amplified gene in malignant human gliomas (70), GLI1 amplification is infrequent in most cancers such as GBM (71). However, since GLI1 is a vital downstream target of the HH pathway, the mRNA level of GLI1 is a reliable indicator of HH pathway activity, this suggests that control of GLI1 protein conversion is critical for GLI-dependent transcription and regulation of the HH signaling pathway (53). And GLI1 protein levels are upregulated in a variety of cancers, and high levels of GLI1 are often associated with tumor progression (72,73). Low GLI1 mRNA levels were similarly negatively correlated with survival in patients with GBM. GLI1 mRNA expression in GBM was significantly lower than in patients with high-HH-medulloblastoma (MB) but significantly higher than in patients with low-HH-MB, and GLI1 mRNA expression is a single continuous distribution, rather than being discrete high/low clusters (5,74). GLI1 promotes the nuclear import of GLI1 into GBM multiforme cells through its transcription factor Forkhead box M1 (FOXM1), thereby increasing the expression of its target genes (75).

Zhou et al (73) reported that in GBM cells, ubiquitin specific peptidase 48 (USP48) gene knockout inhibited cell proliferation and downstream GLI1 target gene expression, thereby inhibiting glioblastoma by USP48 removing ubiquitin-binding compounds on GLI1 and thereby inhibiting GLI1-dependent proteasome degradation. In addition, to a certain extent, GLI1 determines the effect of USP48 on cell proliferation and tumorigenesis and the HH pathway also induces USP48 expression via GLI1 trans-activation, thus forming a mutual feedback loop (73). Similarly, Chang et al (8) indicated that Engraile 1 (EN1) was highly expressed in GBM cells and tissues and positively regulated GLI1 levels. In addition, EN1 also affected HH signal transduction by regulating PC length and the PC transport-related protein TUB-like protein 3, a PC transport-related protein, to control the proliferation, colony formation, migration and tumorigenesis of GBM cells in vivo. Truncated GLI1 (TGLI1) acted as a functionally enhanced GLI1 with an enhanced ability to promote angiogenic heparanase expression. In vivo and in vitro, TGLI1 is more likely to promote GBM angiogenesis and growth than GLI1. Therefore, TGLI1 is a novel mediator promoting GBM angiogenesis through the HH signaling pathway and heparinase is a novel transcriptional target of TGLI1, providing new clues for molecular pathways of tumor angiogenesis and invasive growth (76).

It was observed that both the activation of metabolic glutamate receptor subtype 4 and naringin are able to inhibit the expression of GLI-1 in cells and affect HH signaling pathway transduction, thus inhibiting cell proliferation and promoting cell apoptosis to inhibit the growth of GBM cells (77,78). These may be potential drug targets for controlling GBM cell growth by blocking HH signaling.

Molecular crosstalk is present between mTORC1/2 and HH pathway activity (71,79,80). In GBM, higher mTORC2 activity enhances the expression of several HH pathway molecules (GLI1, GLI2 and PTCH1). A further study by Maiti et al (80) indicated that mTORC2 inhibits GLI2 ubiquitination by inactivating GSK3β, thereby promoting GLI2 stability and nuclear translocation, then modulating the role of HH pathway activity in GBM angiogenesis, metastasis, cell proliferation and cancer stem cell (CSC) regeneration. In addition to influencing mTORC1/2 and HH pathway interactions, GLI2 also affects HH and Wnt pathways and has an important role in GBM stem cell (GSC) maintenance. GLI2 knockdown using lentiviral-mediated shRNA downregulated HH-related and Wnt signaling pathway-related genes, including leucine-rich repeat-containing G-protein coupled receptor 5, inhibited tumor cell proliferation and invasive capacity, and induced apoptosis (81). Takezaki et al (7) indicated that overexpression of GLI2DC, a C-terminal truncated form of GLI2, antagonized GLI transcription factor function, inhibited glioma-initiating cell proliferation in culture and neoplasms occurring in organisms; glioma-initiating cell proliferation was prevented by clipping glial downstream factor cell division cycle 2 (CDC2). These results suggested that the HH/GLI/CDC2 signaling cascade has an important role in glioma-initiating cell proliferation and malignancy. Since GLI2 affects downstream multiple carcinogenic and cancer-inhibiting pathways and is a key player in the network of neoplasmic microenvironments, the possibility of blocking multiple pathways by targeting GLI2 may be a promising strategy.

In vertebrates, two PTCH homologs have been isolated: PTCH1 and PTCH2 (82). PTCH1 is mainly expressed in SHH protein-producing mesenchymal cells, while PTCH2 is expressed in skin and testicular epithelial cells (83). A large clinical cohort study using the TCGA-GBM database detected GLI1 expression in relation to PTCH1. The strong correlation between GLI1 and PTCH1 expression was indicated to be a potential marker of HH-pathway activity (84), since PTCH1 is a true target of GLI1 transcription factors (85) and its expression is expected to increase with the activity of GLI1 (5). Marjanovic Vicentic et al (86) reported increased expression of HH ligand-receptor PTCH and HH effectors GLI1 and GLI2 in U87 and U251 cells overexpressing SOX3. BBF2H7 is an endoplasmic reticulum stranded transmembrane basic leucine zipper transcription factor that binds to HH ligand and PTCH1 to promote the formation of ligand-receptor complexes, thereby activating HH signal transduction (87). Iwamoto et al (88) further indicated that the c-terminal end of secreted lumen BBF2H7 participates in HH ligand-dependent GBM proliferation by binding to HH ligands and PTCH1 to activate HH signaling. Therefore, SOX3 and BBF2H7C terminals may become novel targets for anticancer drug development.

The HH, mTOR, Notch and Wnt/β-catenin signaling pathways are important signaling pathways that regulate GSC stemness and self-renewal (27,89,90), However, the self-renewal and abnormal differentiation of GSCs and their ability to promote the formation of drug resistance to RT and chemotherapy are the main reasons for the recurrence and invasion of GBM after conventional treatment (91,92). The mechanisms of how GSCs during invasion through the HH pathway, particularly in the face of complex and changing brain tissue anatomy, are presented in Fig. 2.

It has been reported that related homolog genes [e.g. Quaking homolog I)] (93), as well as transcription factors (e.g. Nanog homeobox) (94) and sialidase (e.g. neuroaminidase 4 (95) are able to activate the HH signaling pathway to maintain the self-renewal ability of GSCs by increasing SHH/GLI1 expression. This may promote the development of stem-like traits of GSCs and the formation and migration of GBM cell spheres (93–95). In addition, scaffold protein discs large homolog 5 (DLG5) and differentiation inhibitor 1 (ID1) regulate HH pathways by inhibiting downstream target ubiquitination (e.g. GLI1/2) and reducing GLI1/2 degradation (27,96). Cullin-3 interacts with GLI1/2 and dishevelled segment polarity protein 2 and induces their degradation through ubiquitination (27). ID1 and DLG5 inhibit cullin-3 ubiquitin ligase, activate HH signaling and promote GSC proliferation and tumorigenicity (27,96). Park et al (97) demonstrated that dihydro pyrimidine-associated protein 5 (DRP5) is particularly upregulated in the proneural (PN) subtype of GSC and has a key role in maintaining GSC characteristics, including tumor globule formation, stem cell marker expression and xenograft tumor growth, and DRP5 is considered to be a functional biomarker of GBM derived from PN-GSCs. The emergence and maintenance of CSCs are usually controlled by the tumor microenvironment. The tumor microenvironment always provides metabolic challenges to cancer cells and CSCs, mainly due to tissue hypoxia. Mondal et al (98) revealed that nutritional deprivation-induced enhanced the expression of specific biomarkers for GSCs, with higher invasiveness and angiogenic characteristics. These cells induced by microenvironmental nutritional stress (NS) have a high xenoefflux capacity and are therefore resistant to numerous anticancer drugs. The mechanism is that NS activates the Wnt and HH signaling pathways by regulating the β/AKT axis of β-catenin and GLI1, respectively. Vascular endothelial cells in the tumor microenvironment may provide SHH to further activate HH signaling pathways, thereby promoting GSC properties (99).

In summary, the neurobiology and basement membrane invasiveness of neural stem cells involves multiple molecular pathways that are interrelated. Therefore, targeting cross-signaling pathways (e.g., Wnt/HH signaling) and specific markers may be a better therapeutic approach for GSCs.

The mechanisms of GBM cell migration and invasion are complex and involve a series of mechanisms, including adhesion of GBM cells to the extracellular matrix (ECM) and ECM remodeling and degradation (102). As with other malignant tumor types, the growth, metastasis and invasion of GBM also depend on tumor angiogenesis. Although gliomas are characterized by hypervascularization, there are unavoidable disadvantages to anti-angiogenesis, such as reactive resistance mediated by the tumor microenvironment, and invasion and metastasis of tumor cells activated by hypoxia responses (103,104). During invasion and metastasis, GBM cells lose the polarized phenotype of epithelial cells and acquired mesenchymal characteristics, which is referred to as EMT (105). EMT is an active, drug-resistant, low-proliferative transient state that is frequently a feature of cancer as a whole but is seen in GBM in particular (106–108). Tubasatin A, a histone deacetylase 6 (HDAC6) inhibitor, reduced the expression of mesenchymal markers in GBM cells and contributed to the reversal of EMT (109). Feng et al (110) developed a pegylated poly (lactic acid) based nano-drug delivery system (nanoparticles) and modified CK peptides on its surface via GYG connectors to promote multitargeted delivery of Paclitaxel vasculogenic mimicry channels, tumor neovascularization and glioma cells. Similarly, Kast et al (111) proposed the EIS regimen (combination of itraconazole, metformin, naproxen, pirfenidone, quetiapine and rifampicin) that was able to safely and effectively block EMT of GBM. GBM progression may be inhibited by targeting tumor angiogenesis and EMT. Although these animal models are not perfect, they may be used to explore the effectiveness of new treatments for GBM prior to clinical phase I/II studies.

The HH signaling pathway is closely related to PC function, and thus, inhibiting PC function may help inhibit GBM proliferation, malignant development and treatment resistance (112). A previous study reported that the development of resistance to acquired kinase inhibitors is associated with upregulation of PC, uncontrolled PC length and abnormal activation of SHH signaling. Knockdown of KIF7 was observed to control the length and integrity of the PC and re-sensitize GBM cells (113). In addition, Dynarrestin was able to reversely inhibit intraflagellar transport of SMO flux in PC and inhibit HH pathway-dependent neuronal precursors and tumor cell proliferation (23). Therefore, Dynarrestin is a promising compound for the pharmacochemical development of anticancer drugs.

HH signaling has been reported to be abnormally activated in >30% of solid tumor types, including GBM (62,114). Abnormal activation of the SHH pathway is associated with GBM resistance to temozolomide (TMZ) and the reason is the high expression of methylguanine methyltransferase (MGMT), which reverses the effects of TMZ on DNA (115,116) and confirms cell protection from TMZ-induced death by silencing three genes: MutS homolog 2 (a DNA repair protein involved in MMR), PTCH2 and chloride channel accessory 2 (a type 1 transmembrane protein that inhibits the Wnt pathway) (24). Resistance to TMZ was only slightly reversed by MGMT inhibitor O6-benzylamine, but a marked further enhancement was achieved by addition of Honokiol (117). Furthermore, the invasion of GBM was reported to be associated with the presence of CSCs and the SHH pathway has an important role in the maintenance and proliferation of CSCs (118,119). After inhibiting SHH, LDE225 slowed down the growth of GBM and downregulated PTCH1 and GLI1 in vivo (120). CSCs preferentially activate the DNA damage checkpoint response and exhibit enhanced DNA repair ability; thus, SHH signaling via GLI1 in CSC has a role in GBM resistance to TMZ (121).

Glabrescione B is the first small molecule to bind to GLI1 zinc fingers, impelling GLI1 activity by interfering with its interaction with DNA. Thus, it inhibits the ability of HH-dependent tumor stem cells to self-renew and cladogenesis. The determination of the structural requirements for GLI1/DNA interactions highlights their relevance to drug interference with GLI signaling (10). Melamed et al (122) developed polyethylene imine-coated spherical nucleic acid nanoparticles (PEI-SNAs) targeting GLI1. GLI1 PEI-SNAs bind scavenger receptors on GBM cells and undergo endocytosis in a pit/lipid raft/dynein-dependent manner, promoting the silencing of HH pathway genes and downstream target genes. These genes promote an aggressive, drug-resistant GBM phenotype. GLI1 PEI-SNAs not only significantly increased the sensitivity of nerve spheres to chemotherapy, but also further impaired the formation of dry nerve spheres (123). Arsenic trioxide also significantly reduced the cladogenesis of tumor neuroglobules by inhibiting the HH pathway, inhibiting the proliferation of GBM neuroglobules and promoting apoptosis (124). The combination of the SHH inhibitor GANT-61 with TMZ increased the cytotoxicic effect of TMZ and the combination of GANT-61 with TMZ increased the production of reactive oxygen species in GBM cells, suggesting that inhibition of the SHH pathway may sensitize GBM cells to the effects of TMZ by increasing oxidative stress (114,124). GANT-61 induced apoptosis and autophagy in GBM cells by increasing the expression of light chain 3II and lysed Caspase-3 and −9 (125). Furthermore, GLI inhibition combined with TMZ increased the apoptosis rate of glioma stem cells by 6.8-fold, thereby reducing the size and number of nerve spheres grown from glioma stem cells (115). Yin et al (126) reported that the combination of tumor suppressor gene miR-326 and curcumin significantly inhibited the SHH/GLI1 pathway of glioma cells, independent of the P53 status, significantly increased apoptosis and reduced the proliferation and migration of glioma cells. Similarly, Ji et al (127) reported that a novel PI3K inhibitor, XH30, inhibited tumor growth that was resistant to TMZ. In terms of the mechanism, the role of XH30 may be to reverse the activation of GLI1 induced by SHH by atypical HH signaling and to reduce GLI1 activation by insulin-like growth factor 1 (127). Thus, XH30 may be a novel treatment option for TMZ-resistant GBM.

Traditional treatments for GBM include systemic chemotherapy, RT and surgery. Han et al (128) synthesized three phosphorylated peptides derived from GLI2 and combined them with the cell-penetrating peptide TAT-[47–57]AYGRKKRRQRRR. The three mixed phosphorylated polypeptides derived from GLI2 significantly increased the level of GLI2 phosphorylation and decreased the transcriptional activation of GLI2, and the radiation sensitization of GBM cells was significantly higher than that in the control group (128). HDAC6 was upregulated in GSCs and inhibited HDAC6 down-regulated glioma-associated oncogene GLI1 and PTCH1/2 receptors, as well as SHH signaling components, expression and activity, thereby inhibiting GSC proliferation, inducing differentiation and increasing the apoptosis rate through the SHH/GLI1 signaling pathway (109,129). Inhibition of HDAC6 by Tubasatin A enhanced the radiosensitivity of GBM tumor cells. The mechanism may be that HDAC6 inhibits checkpoint kinase (CHK)1 degradation induced by down-regulation of X-linked inhibitor of apoptosis, which reduced the DNA damage repair ability of GSCs, leading to increased radiosensitivity (109,130).

In summary, target genes associated with the SHH/GLI pathway provide promising new drug targets for inhibiting GBM proliferation, as well as overcoming drug resistance and radiation resistance of GSCs.

The steroidal alkaloid cyclopamine, an antagonist of the HH coreceptor SMO, acts as an inhibitor of the HH pathway (131). To limit the toxicity of cyclodopamine to HH-dependent non-tumor cells, cyclodopamine precursor drugs [e.g., cyclodopamine glucuronoside precursor drugs (CGP-2) and 1b] are commonly used (132,133). It was indicated that CGP-2 inhibits the HH pathway more effectively than conventional TMZ adjuvants (131). In the presence of β-glucuronidase, the activated prodrug 1b was toxic and downregulated the HH target gene GLI1 in C6 cells and C6-CSCs (132). In U251 cells, tyramine not only inhibited the HH/GLI1 signal transduction pathway, leading to decreased MGMT expression, but also induced cell apoptosis by activating caspase-3 cleavage, thus leading to increased sensitivity of GBM to TMZ (133). However, the combination of acepromazine and TMZ enhanced the dryness and drug resistance of GBM cells by inducing the expression of SOX-2 and OCT-4 and may lead to tumor recurrence in patients (134). Therefore, the best therapeutic strategy is to first inhibit the SHH pathway and then administer TMZ (134,135). Liu et al (136) found that the combination of capsulated propylamine and TMZ had synergistic cytotoxic effects and was more likely to inhibit the ability to induce apoptosis and eliminate neuroglobin formation by inhibiting GLI1 expression. Therefore, MCyp may be used as a tumor stem cell inhibitor to prevent tumor recurrence. Future efforts should be made to investigate the possibility of using HH pathway inhibitors prior to conventional chemotherapy in patients with GBM. Future efforts should focus on the efficacy of HH pathway inhibitors prior to systemic chemotherapy in patients with GBM.

Chen et al (137) indicated that PF403 inhibits cell surface Smoothened (Smo) receptor aggregation at the molecular level by directly binding or enhancing the interaction between Smo and the suppressor PTCH1. In addition, PF403 significantly inhibited the transcription of GLI1 and its accumulation in the nucleus by promoting the interaction between SUFU-GLI1 and PKA-GLI1, blocking the HH signaling pathway of T98G MGMT-expressing cells, and downregulated the expression of MGMT. Inhibition of the HH pathway by PF403 counteracted TMZ resistance and the precursor Cat3 of PF403 enhanced the anti-tumor activity of TMZ in vivo (137,138). In summary, Cat3 is a promising therapeutic agent for HH-driven GBM.