Non-coding RNAs and glioma: Focus on cancer stem cells

Abstract

Glioblastoma and gliomas can have a wide range of histopathologic subtypes. These heterogeneous histologic phenotypes originate from tumor cells with the distinct functions of tumorigenesis and self-renewal, called glioma stem cells (GSCs). GSCs are characterized based on multi-layered epigenetic mechanisms, which control the expression of many genes. This epigenetic regulatory mechanism is often based on functional non-coding RNAs (ncRNAs). ncRNAs have become increasingly important in the pathogenesis of human cancer and work as oncogenes or tumor suppressors to regulate carcinogenesis and progression. These RNAs by being involved in chromatin remodeling and modification, transcriptional regulation, and alternative splicing of pre-mRNA, as well as mRNA stability and protein translation, play a key role in tumor development and progression. Numerous studies have been performed to try to understand the dysregulation pattern of these ncRNAs in tumors and cancer stem cells (CSCs), which show robust differentiation and self-regeneration capacity. This review provides recent findings on the role of ncRNAs in glioma development and progression, particularly their effects on CSCs, thus accelerating the clinical implementation of ncRNAs as promising tumor biomarkers and therapeutic targets.

Graphical abstract

The presence of cancer stem cells in a tumor causes metastasis to spread more readily. NcRNAs can regulate cellular signaling in CSCs and glioma cells. However, more research into the exact pathways and mechanisms of action of ncRNAs in CSCs and glioma, is required to develop more effective therapy for glioma.

Introduction

Gliomas are the most common primary type of adult brain cancer, consisting of up to 80% of malignant brain tumors.¹ While being the most frequent primary brain tumor, glioblastoma (GBM) is a type of glioma accounting for 57.3% of these tumors and has the worst prognosis: WHO grade IV.^1,2 Gliomas are divided into two distinct categories. Firstly, the IDH wild-type tumor or de novo primary GBM, which is most commonly found in older patients (≥62 years), accounts for about 90% of all GBMs. Secondly, the IDH mutant type or secondary GBM, which more frequently occurs in 40- to 50-year-old patients and accounts for only 10% of cases. IDH mutant tumors arise from underlying low-grade astrocytomas.^2,3

Microarrays and next-generation sequencing technology have led to significant advances in whole-genome sequencing and provided a more comprehensive understanding of non-coding RNAs (ncRNAs) and their roles and functions. The majority of the human genome (>90%) undergoes transcription, but many of these genes do not result in the synthesis of new proteins.⁴ Several ncRNAs have important regulatory functions. ncRNAs, including lncRNAs (long ncRNAs), miRs (microRNAs), and circRNAs (circular RNAs) play critical roles in numerous cell processes and are regulated by specific molecular mechanisms.⁵,⁶ miRs are a group of short endogenous ncRNAs that regulate the post-transcriptional expression of many genes.⁷ miRs are involved in many pathological and physiological cellular processes, including tumorigenesis and cancer progression. Dysregulated miRNA expression may result either in tumor inhibition or in tumor promotion as an oncogene.^8,9 circRNAs are a class of ncRNA with covalently closed loops and high stability. Growing evidence has shown that circRNAs play critical roles in the development and progression of diseases, particularly in cancer growth, metastasis, stemness, and resistance to therapy.¹⁰ lncRNAs are a group of functional ncRNAs with a wide range of major regulatory functions in proliferation and differentiation, as well as tumor progression or tumor suppression.^6,11, 12, 13, 14 Here, we review the current information on the role of ncRNAs in glioma, particularly their effects on cancer stem cells (CSCs).

CSCs and glioma

The cellular heterogeneity in CNS tumors has long been appreciated^15,16; however, the role of self-regenerating tumor cells with increased tumorigenesis has been poorly recognized. Up to now, different terms have been used to denote these cells, including tumor/cancer/brain stem cells, stem-like tumor cells, tumor/cancer/glioma/brain tumor-propagating cells, and glioma/cancer/brain tumor-initiating cells. Because of these inconsistencies, attention has shifted away from their biology and their role in tumorigenesis, toward the discovery of new markers expressed on these cells, and determining if these cells can replicate as floating (non-adherent) spheroids. Moreover, these tumor cells are not necessarily produced from transformed stem cells, and other cell types, including normal stem cells and well-differentiated progenitor cells, could undergo oncogenic transformation. Therefore, precise functional assays must be performed, and an accepted definition should be used in all experimental studies. Any population of CSCs must have the capacity for self-regeneration, and also be able to produce well-differentiated progeny (Figure 1). In the case of brain tumors, these cells can form a tumor following intracranial transplantation, recapitulating the heterogeneity of parental tumor cells. Tumor-initiating cells in animal models can be used for investigation, but CSCs are more infiltrative capability than their progeny, and also their progeny lose tumorigenic potential during differentiation. The presence of a cellular hierarchy can be demonstrated by prospective enrichment and depletion of tumorigenic and non-tumorigenic cells. Cancer cells that contain a cellular hierarchy and are tumorigenic, are considered glioma stem cells or glioma CSCs. Cell culture spheroids can be derived from brain cells (normal or neoplastic), and their progenitors have limited self-renewal potential. However, the mere ability to form spheroids does not define CSCs, without showing a self-renewing population.¹⁸ High-passage cell lines are unlikely to be functionally validated CSC models, and cannot accurately represent tumor complexity in vivo.¹⁹

Figure 1.

Functional criteria of CSCs

(A) CSCs are defined by functional characteristics that include sustained self-renewal, persistent proliferation, and tumor initiation upon intracranial transplantation, which is the definitive functional CSC assay. (B) CSCs also share features with somatic stem cells, including frequency within a tissue or tumor, stem cell marker expression (examples relevant to GBM and the brain are provided), and the ability to generate progeny with multiple lineages. Bmi1, B cell-specific Moloney leukemia virus insertion site 1; Olig2, oligodendrocyte transcription factor 2; Sox2, SRY-box transcription factor 2. This figure was adapted from Lathia et al.¹⁷

At high passage numbers, cell lines exhibit changes in morphology, reduced or altered key functions and efficiency, and frequently no longer represent reliable models of their original source material due to selective pressures and genetic drift. Cancer cell lines have significant limitations due to a lack of vascular, stromal, and immune components. Tumors are ecosystems of evolving clones that compete or cooperate with each other and other normal cells that infiltrate their microenvironment.²⁰ This begs the interesting question of whether these clones were selected during their growth into the culture medium or through cell passaging over time. As a result, cell lines derived from a single clone are not always representative of the diversity present in the original tumor.²¹

Thus, whereas the growth of glioma cells as neurospheres is not essential to retaining stemness, the microenvironment, including medium composition and culture conditions, influences the CSC properties.^18,22,23

After the adoption of CD133 as the first surface marker for GCSs, they were classified as CD133⁺ and CD133⁻. CD133⁺ cells or CSCs gradually lose their ability to self-renew during differentiation, but CD133 expression allows brain tumors to form in vivo, and neurospheres to grow in vitro.^22,24, 25, 26, 27 Although other surface markers have been reported, which could be used to classify GSCs, the most useful marker remains CD133.²⁸ Prognostic indicators for GBM progression, include CD133^+/Ki-67⁺ cells, and the expression of HOX or Nestin genes.29, 30, 31 CD184 (CXCR4 chemokine-receptor) is another surface marker that is significantly correlated with CD133⁺ cells and has been shown to increase the expression of hypoxia-inducible factor 1 (HIF-1).^32,33 Another surface marker is MUSASHI-1, which regulates the cell cycle and is an RNA binding protein involved in post-transcriptional gene editing.³⁴ Many additional surface markers have been described that might be used to identify GSCs, such as the cell surface gangliosides GFAP, KLF4, SALL4, ALDH1, L1CAM, SOX2, CD90, and A2B5, and the cell surface glycoprotein CD44.34, 35, 36, 37, 38, 39, 40 Although CD133 is a cell surface marker that enriches GSCs, the use of CD133 as a unique glioma stem cell marker is likely not enough to tag the whole self-renewing cancer cell reservoir and additional research is needed to identify more markers for GBM stem cells.41, 42, 43 The main biomarkers of glioma stem cells are illustrated in Figure 2.

Figure 2.

Schematic overview of the cellular components of the microenvironment of GBM

The tumor microenvironment is a complex network composed of stromal cells (fibroblasts, microglia, astrocytes), mesenchymal cells, stem cells, and immune and inflammatory cells (macrophages). The main biomarkers of GSCs are indicated. This figure was adapted from Alves et al.⁴⁴

miRNAs and CSCs in glioma

Members of the miR-17 family, miR-20a and miR-106a are expressed in multiple types of cells.⁴⁵ Both upregulation and downregulation of miRNAs have been observed in various cancers, and specific miRNAs may either promote or suppress tumor formation. 46, 47, 48, 49, 50 miRNAs have also been implicated in the function of stem cells (both normal and cancer). For instance, upregulation of the miR-17-92 cluster (including miR-20a) induces pulmonary epithelium progenitor cells to proliferate and prevents them from differentiating.⁵¹ In mouse embryos, miR-20a/106a acts to control stem cell differentiation.⁵² miRNAs are also present in high levels in MLL leukemia stem cells and could affect their function by regulating p21.⁵³ The anti-tumor activity of tissue inhibitor of metalloproteinases-2 (TIMP-2) has been reported in CSCs in various studies.54, 55, 56

miR-106a, one of the tumor-suppressor miRNAs, has played a significant role in the development and progression of human tumors. It was upregulated in colorectal cancer,⁵⁷ gastric carcinoma,⁵⁸ and mantle cell lymphoma,⁵⁹ whereas it was downregulated in glioma. In their study, Dai et al. found that overexpression of miR-106a downregulated expression of glucose transporter 3 (GLUT3) or SLC2A3, an oncogene in several human cancers, via targeting 3′ UTR of SLC2A3, resulted in suppression of cell proliferation and cell glucose uptake in GBM cells.⁶⁰ miR-20a is widely upregulated in diverse cancer subtypes, including hepatocellular cancer, lung cancer, and GBM.

The bioinformatic analysis confirmed using a luciferase reporter assay showed that miR-20a can directly and negatively regulate CELF2 (CUGBP Elav-like family member 2) gene expression, thus playing a critical role in the growth and invasion of glioma cells.⁶¹ In addition, miR-20a may regulate cell invasion of GBM via IL-6/JAK2/STAT3 axis belonging to the JAK/STAT signaling pathway.⁶² Xu et al. suggested that miR-20a introduces its oncogenic activity by the HIF-1a/c-MYC pathway in IDH1 R132H-mutant glioma.⁶³ The presence of this mutation in glioma upregulates HIF-1a expression, which decreases c-MYC activity, resulting in a consequential decline in miR-20a, is responsible for glioma cell proliferation and resistance to temozolomide (TMZ) treatment.⁶³

TIMP-2 is another target gene of miR-20a/106a, which can deregulate the expression of the TIMP-2 gene by interfering with the 3′ UTR of TIMP-2 mRNA in GBM.

Nordy (dl-nordihydroguaiaretic acid) is a small-molecule lipoxygenase inhibitor,⁶⁴ which has been found to suppress cancer growth.65, 66, 67 In vitro as well as in vivo studies in glioma have shown its ability to modulate differentiation and inhibit growth.⁶⁶ It was proposed that Nordy could drive GSCs toward differentiation,⁶⁵ by increasing the expression of TIMP-2 via downregulation of miR-20a and miR-106a. Wang et al. examined the ability of miR-20a/106a to promote the invasion of GSCs.⁴⁵ Compared with regular glioma cells, GSCs had higher expression of miR-20a/106a, which was associated with their invasion. miR-20a/106a was found to target TIMP-2 expression, which showed a negative correlation with the levels of these miRs. miR-20a/106a downregulation resulted in an increase in TIMP-2, which suppressed GSC invasion. The ability to suppress miR-20a/106a and therefore upregulate TIMP-2 was proposed to explain the anti-tumor effect of Nordy.⁴⁵

Both anti-cancer and pro-oncogenic effects of miR-146a have been reported,68, 69, 70, 71, 72 and it has also been shown to be associated with prolonged survival in GBM patients.⁷³

POU3F2 is a transcription factor belonging to the POU-domain transcription factor and is a key differentiation factor in neurons and embryonic development.⁷⁴ POU3F2 knockdown can cause tumor-suppressor effects in various cancers.75, 76, 77, 78 It can be targeted to prevent prostate tumors from neuroendocrine differentiation.⁷⁸ CircPOLR2A, an upregulated circRNA in GBM cells, activates the transcription of Sox9 through the miR-2113/POU3F2 axis, thus enhancing GBM cell growth.⁷⁹ POU3F3 modulates cell proliferation by G1 cell-cycle arrest and apoptosis via its influence on DLL1 and Sox2. One study showed that long intergenic ncRNA POU3F3 (linc-POU3F3) is overexpressed in high-grade glioma tissues and promotes cell viability and proliferation of glioma cells, and leads to glioma progression through downregulation of POU3F3.⁸⁰ In a recently published paper by Yang et al.,⁸¹ they reported that overexpression of POU2F2 significantly correlated with poor prognosis of GBM patients.^81,82 They indicated that POU2F2 induces a metabolic shift toward aerobic glycolysis and promotes cell growth and GBM progression through PDPK1-dependent activation of the PI3K/AKT/mTOR pathway.^81,82

SMARCA5 is a member of the SWI/SNF family, with helicase and ATPase properties. SMARCA5 can enhance cancer development in ovarian and glioma tumors.^83,84 The suppressive effects of miR-100 on breast cancer stem cells could be partly mediated by regulating SMARCA5.⁸⁵ Interestingly, the level of miR-146a was negatively associated with SMARCA5 in bladder cancer.⁸⁶ Cui et al. investigated the role of miR-146a and its downstream pathways in GBM.⁸² They reported a significant downregulation of miR-146a as a result of promoter hypermethylation in recurrent GBM patients, which was associated with a poor prognosis. In vitro as well as in vivo findings showed that miR-146a upregulation greatly reduced proliferation and invasion, as well as the stemness of glioma cells, and enhanced their TMZ sensitivity. At the molecular level, these effects suggest the ability of miR-146a to suppress POU3F2 and SMARCA5 by directly targeting their 3′ UTRs in GBM cells. Altogether, their results suggest that miR-146a may inhibit stemness properties in GBM cells, and enhance their TMZ sensitivity.⁸²

miR16 could suppress invasion and migration in glioma cells.87, 88, 89 In GBM cells, miR16 suppressed invasion, adhesion, and downregulated genes involved in epithelial-mesenchymal transition (EMT).⁹⁰ Since an association between SOX2 and stemness has been reported, investigating the interplay between miR16 and the SOX family transcription factor in GSCs may reveal useful information. SOX family members SOXD and SOXE have been implicated in glioma formation.⁹¹ The transcription factor SOX5 acts to maintain the chromatin configuration and regulates gene expression in various developmental pathways. SOX5 was able to suppress proliferation in glioma cells.^92,93 Tian et al. examined the levels of miR16 in GBM SGH44, U87, and U251 cells, and in GSCs, how it affected tumor progression, and its role as a possible prognostic marker.⁹⁴ Both in vitro and in vivo, upregulation of miR16 suppressed tumor progression, while its inhibition was associated with tumor promotion. There was a positive correlation between miR16 levels and GSCs’ ability to differentiate, and a negative correlation with migration, invasion, and the ability to form colonies. Bcl2, CCND1, CCNE1, CDK6, and SOX5 were identified as direct targets of miR16, and all these factors were downregulated by miR16 in cells. Finally, they showed a correlation between miR16 levels and clinical outcomes in GBM patients and suggested that the anti-cancer effect of miR16 involved GSCs.⁹⁴

miR-182 was first detected in murine neurosensory tissues.95, 96, 97 Although miR-182 levels are scarce in the fetal period, it becomes upregulated after birth, where it can induce the retinal progenitor cells to terminally differentiate, as well as maintain their mature form.⁹⁶ It may also induce differentiation and the mesenchymal-to-epithelial transition by modulating SNAI2.⁹⁸ c-Met has been reported to be upregulated in GBM,99, 100, 101 where it promotes tumor invasion and proliferation.^102,103

Hypoxia-inducible factor 2α (HIF2A) is secreted in hypoxic conditions to enhance GSC survival and proliferation, and there was a negative correlation between HIF2A levels and glioma prognosis.¹⁰⁴ Another pro-oncogenic factor, Bcl2-like12 (Bcl2L12), was reported to be upregulated in GBM.^105,106 Kouri et al. examined the role of miR-182 in GBM, and whether it could be a prognostic marker.¹⁰⁷ They identified that miR-182 is a tumor suppressor, which could inhibit Bcl2L12, c-Met, and HIF2A, and subsequently could prevent GSC growth and stemness, and possibly improve GBM treatment response. The same results were observed in vivo. They demonstrated an association between miR-182 and GBM prognosis and suggested that miR-182 could suppress GSCs by inhibiting Bcl2L12, c-Met, and HIF2A.¹⁰⁷

miR-302-367 has been shown to play an important role in mesendoderm differentiation.¹⁰⁸ In addition, miR-302-367 may also regulate stemness properties in stem cells of different origins.^109,110 It was shown that stemness transcription factors, such as Oct4, Sox2, and Nanog, could regulate miR-302-367 in ESCs, as well as early development of murine cells.^109,111 SDF1 can bind to CXCR4 to regulate various signaling pathways, such as PLC, PI3K/AKT, and MAPK, and to affect multiple cellular processes.¹¹² CXCR4 is involved in proliferation and motility and could promote the aggressive phenotype of glioma, which explains its correlation with poor patient prognosis.^33,113,114 In GSCs, the SHH-GLI-NANOG axis was shown to regulate proliferation and stemness.115, 116, 117, 118 Fareh et al. investigated the role of the miR-302-367 cluster in GSCs.¹¹⁹ They used serum to suppress stemness in GSCs, and observed upregulation of the expression of this cluster. They found that miR-302-367 upregulation could inhibit the stemness and tumorigenicity of GSCs by suppressing CXCR4 and disrupting the SHH-GLI-NANOG pathway. They concluded that the miR-302-367 cluster suppressed GSC stemness and tumorigenicity by inhibiting CXCR4 and interfering with the SHH-GLI-NANOG pathway.¹¹⁹

Table 1 lists some microRNAs that have been reported to be involved in CSCs, and GSCs in particular.

Table 1.

Role of microRNAs in cancer stem cells

microRNA	Expression	Target	Model (in vitro, in vivo, human)	Ref.
miR-26a	↑	AP-2α	in vitro, in vivo	Huang et al.120
miR-93	↑ (higher upregulation in PN GSCs than in MES GSCs)	BECN1/Beclin 1, ATG5, ATG4B, and SQSTM1/p62	in vitro, in vivo, human	Huang et al.121
miR-3940-5p	↓	CUL7, NF-κB	in vitro	Xu et al.122
miR-9-5p	↑	NAP1L1, FREM2	in vitro	Zottel et al.123
miR-124-3p	↑	SPRY1, NAP1L1, VIM	in vitro	Zottel et al.123
miR-21-5p	↓	SPRY1	in vitro	Zottel et al.123
miR-138-5p	↓	VIM	in vitro	Zottel et al.123
miR-1-3p	↓	NCL	in vitro	Zottel et al.123
miR-30a	↓	NT5E/Akt signaling pathway	in vitro, in vivo	Peng et al.124
miR-150-5p	↓	Wnt/β-catenin pathway	in vitro, in vivo	Tian et al.125
miR-100-5p	↓	SMARCA5, SMRT, AKT, ERK, ErbB3, ErbB2, p21	in vitro, in vivo	Alrfaei et al.126
miR-146b-5p	↓	SMARCA5, TGF-β pathway	in vitro, in vivo	Wang et al.127
miR-18a	↑	RORA, TNF-α, NF-κB	in vitro, in vivo	Jiang et al.128
miR-145	↓	SOX2-Wnt/β-catenin pathway	in vitro, in vivo	Qian et al.129
miR-29a	↓	PDGFC, PDGFA	in vitro, in vivo	Yang et al.130
miR-320a	↓	PHF6, MMP16, MCL1	in vitro, in vivo	Kang et al.131
miR-4496	↓	ABCB1, ABCG2, MGMT, PHF6,MMP16, MCL1	in vitro, in vivo	Kang et al.131
miR-26a	↑	PTEN, PI3K/Akt	in vitro, in vivo	Wang et al.132
miR-504	↓	Grb10	in vitro, in vivo	Bier et al.133
miR-486-5p	↑	PTEN, FoxO1	in vitro, in vivo	Lopez-Bertoni et al.134
miR-1300	↓	ECT2	in vitro	Boissinot et al.135
miR-603	↓	IGF1, IGF1R	in vitro, in vivo	Ramakrishnan et al.136
miR-200b	↓	CD133/PI3K/Akt signaling axis	in vitro	Liu et al.137
miR-107	↓	Notch2, MMP-12	in vitro, in vivo	Yuan et al.138
miR-302-367	↓	CXCR4/SDF1, SHH, cyclin D, cyclin A, E2F1	in vitro, in vivo	Fareh et al.139
miR-370-3p	↓	NEAT1, HMGA2, HIF1A	in vitro, in vivo	Lulli et al.140
miR-141	↓	Jagged1	in vitro, in vivo	Gao et al.141
miR-7-5p	↓	Yin Yang 1	in vitro	Jia et al.142
miR-33a	↑	PDE8A→PKA UVRAG→NOTCH	in vitro, in vivo	Wang et al., 2014143
miR-203	↓	–	in vitro	Deng et al.144
miR-128	↓	BMI1, SUZ12	in vitro	Peruzzi et al.145
miR-145	↓	ABCG2	in vitro	Shi et al.146
miR-148a	↑	GADD45A	in vitro, in vivo	Cui et al.147
miR-30	↑	Jak/STAT3, SOCS3	in vitro, in vivo	Che et al.148
miR-205	↓	–	in vitro, in vivo	Huynh et al.149
miR-300	↑	LZTS2	in vitro, in vivo	Zhang et al.150
miR-143	↓	hexokinase 2	in vitro, in vivo	Zhao et al.151
miR-10b	↑	–	in vitro, in vivo	Guessous et al.152
miR-146b-5p	↓	HuR/lincRNA-p21/β-catenin axis	in vitro, in vivo	Yang et al.153
miR-124	↓	KITL, SEMA6D, NRP2, THBS1	in vitro, in vivo	Marisetty et al.154
miR-203	↓	KITL, SEMA6D, NRP2, THBS1	in vitro, in vivo	Marisetty et al.154
miR-34a	↓	cyclin D1, c-myc, c-met, Ki-67 Bcl-2 Family	in vitro	Sun et al.155
miR-30b-3p	↑	RHOB	in vitro, in vivo	Yin et al.156
miR-135a	↓	Arhgef6	in vitro, in vivo	Hemmesi et al.157
miR-138	↑	CASP3, BLCAP, MXD1	in vitro, in vivo	Chan et al.158
miR-153	↓	Dvl-3	in vitro	Zhao et al., 159
miR-146a	↑	NUMB	in vitro	Puca et al., 160
miR-608	↓	MIF	in vitro	Wang et al., 161
miR-10b	↑	P21, P16, BIM, PTBP2	in vitro, in vivo	El Fatimy et al., 162
miR340	↓	PLAT	in vitro, in vivo	Yamashita et al., 163
miR-34a	↓	c-Met, Notch	in vitro	Guessous et al.91
miR-20a/106a	↑	TIMP-2	in vitro, in vivo	Wang et al., 45
miR-21	↑	FASLG	in vitro	Shang et al., 80
miR-135b	↓	ADAM12, SMAD5, GSK3β	in vitro, in vivo	Lulli et al., 164
miR-223	↑	PAX6, PI3K/Akt	in vitro	Huang et al., 165
miR-153	↓	Nrf-2/GPx1/ROS axis	in vitro, in vivo	Yang et al., 166
miR-125b	↓	E2F2	in vitro	Wu et al., 167
miR-146a	↓	POU3F2, SMARCA5	in vitro, in vivo, human	Cui et al., 82
miR-451	↓	–	in vitro	Gal et al., 168
miR-124	↓	STAT3	in vitro, in vivo	Wei et al., 169
miR-134b	↓	MMP-12	in vitro, in vivo	Liu et al., 170
miR-218	↓	Bmi1, Wnt	in vitro, in vivo	Tu et al., 171
miR-23b	↓	HMGA2	in vitro	Geng et al., 172
miR-296-5p	↓	HMGA1, Sox2	in vitro, in vivo	Lopez-Bertoni et al., 173
miR-125b-2	↑	Bax, Bcl-2, cytochrome c, Apaf-1, caspase-3, PARP	in vitro	Shi et al., 174
miR-198	↑	NNAT	in vitro	Liu et al., 175
miRNA-155-5p	↑	BMP	in vitro	Liu et al., 175
miRNA-124-3p	↓	–	in vitro	Liu et al., 175
miR-455-3p	↑	Smad2	in vitro, human	Tezcan et al., 176
miR-181b	↓	–	in vitro, human	Tezcan et al., 176
miR-125b	↑	Bak1	in vitro, in vivo	Chen et al., 177
miR-137	↓	RTVP-1	in vitro	Bier et al., 178
Let-7b	↓	E2F2	in vitro	Song et al., 179
miR-145	↓	CTGF, SPARC	in vitro, human	Lee et al., 180
miR-92a-3p	↑	CDH1/β-catenin, Notch1/Akt	in vitro, in vivo	Song et al., 181
miR128-1	↓	BMI1, E2F3	in vitro, in vivo	Shan et al., 182
miR-148a	↑	MIG6, BIM	in vitro, in vivo, human	Kim et al., 183
miR-125b	↑	MMP9	in vitro, in vivo	Wan et al., 184
miR-181b	↓	–	in vitro	Li et al., 185
miR16	↓	Bcl2, CDK6, CCND1, CCNE1, SOX5	in vitro, in vivo	Tian et al., 94
miR-182	↓	Bcl2L12, c-Met, HIF2A	in vitro, in vivo, human	Kouri et al., 107
miR-152	↓	KLF4, LGALS3, MEK1/2, PI3K	in vitro, in vivo	Ma et al., 37
miR-29a	↓	QKI-6/WTAP	in vitro, in vivo	Xi et al., 186
miR-125b	↓	CDK6, CDC25A	in vitro	Shi et al., 187
miR-449a	↑/↓	Ccnd1, Gpr158	in vitro, in vivo, human	Li et al., 188
miR-7	↓	EGFR, Akt, NF-κB, DR5	in vitro, in vivo	Bhere et al., 189
miR-9/9∗	↑	CAMTA1, NPPA	in vitro, in vivo, human	Schraivogel et al., 190
miR-18a∗	↑	ERK, DLL3, NOTCH, SHH-GLI-NANOG axis	in vitro, in vivo	Turchi et al., 191
miR-145	↓	NEDD9	in vitro, in vivo, human	Speranza et al., 192
miR-128	↓	Bmi-1	in vitro, in vivo	Godlewski et al., 193
miR-145	↓	JAM-A	in vitro, human	Alvarado et al., 194
miR-126	↑	–	in vitro, in vivo	Halle et al., 195
miR-137	↑	–	in vitro, in vivo	Halle et al., 195
miR-128	↑	–	in vitro, in vivo	Halle et al., 195
miR-218-5p	↓	MMP9	in vitro	Wu et al., 196
miR-128	↓	EGFR, PDGFRα	in vitro, in vivo	Papagiannakopoulos et al., 197
miR-374a	↓	NRN1, CCND1, CDK4, Ki67	in vitro, in vivo	Pan et al., 198
miR-125b	↓	Lin28	in vitro	Wan et al., 199
miR-491	↓	IGFBP2, EGFR, CDK6, Bcl-xL	in vitro, in vivo	Li et al., 200
miR-101	↓	KLF6, CHI3L1, MEK1/2, PI3K	in vitro, in vivo	Yao et al., 201
miR-139	↓	PDE2A, FZD3, β-catenin, GSK-3β	in vitro, in vivo, human	Li et al., 202
miR-154	↑	PRPS1	in vitro, in vivo	Yang et al., 203
miR-124	↓	CDK6, pSer 807/811	in vitro	Silber et al., 204
miR-137	↓	CDK6, pSer 807/811	in vitro	Silber et al., 204
miR-330	↑	SH3GL2, ERK, PI3K/AKT	in vitro, in vivo	Yao et al., 205
miR-302-367 cluster	↓	CXCR4/SDF1, SHH/GLI/NANOG	in vitro, in vivo	Fareh et al., 119
miR-363	↑	caspase-3, caspase-9, Bim	in vitro	Floyd et al., 206
miR-582-5p	↑	caspase-3, caspase-9, Bim	in vitro	Floyd et al., 206
miR-145	↓	Sox9, ADD3	in vitro, in vivo	Rani et al., 207
miR-10	↑	CSMD1, HOXD10	in vitro	Lang et al., 208
miR-124	↓	NRAS, PIM3	in vitro	Lang et al., 208
miR-125b	↑	PIAS3	in vitro, in vivo	Shi et al., 209
miR-134	↓	KRAS, STAT5B	in vitro, in vivo	Zhang et al., 210
miR142-3p	↓	IL-6, HMGA2, Sox2	in vitro, in vivo, human	Chiou et al., 211
CMV70-3P	↑	SOX2	in vitro	Ulasov et al., 212
miR-148a	↓	–	in vitro, in vivo	Lopez-Bertoni et al., 213
miR-124	↓	PTBP1, ANXA7	in vitro	Ferrarese et al., 214
mir34c	↓	Bcl2, NUMB, p73, AKT	in vitro	Iannolo et al., 215
miR-944	↓	VEGFC, AKT/ERK	in vitro, in vivo	Jiang et al., 216
miR-34a	↓	Rictor, PI3K/AKT, WNT/β-Catenin	in vitro, in vivo	Rathod et al., 217
miR-425-5p	↑	FOXJ3, RAB31	in vitro, in vivo	De La Rocha et al., 218
miR-451	↓	AMPK/OCT1/miR-451/LKB1	in vitro, in vivo	Ogawa et al., 219
miR-24-3p	↑	BNIP3	in vitro, in vivo	Zhang et al., 220

lncRNAs and CSCs in glioma

lncRNAs, through several mechanisms, are involved in metabolic reprogramming, cell proliferation, cell apoptosis, cell metastasis and invasion, cell-cycle and genomic instability, EMT and migration, cancer stemness, and drug resistance (Figure 3).²²¹ The oncogenic function of the lncRNA NEAT1 has been shown in glioma and other tumors.222, 223, 224 Low levels of Let-7g-5p (a let-7 family member) have been reported in glioma patient samples,²²⁵ and higher levels may be predictive of better clinical outcomes in GBM.²²⁶

Figure 3.

The role of lncRNAs in regulating cancer cellular processes

Roles of oncogenic versus tumor suppressive lncRNAs and their mechanisms in tumorigenesis.

MAP3K1 has pro-oncogenic effects in glioma, gastric, and breast cancer by regulating proliferation and migration, as well as promoting tolerance to therapy.227, 228, 229 Bi et al. investigated the level and function of NEAT1 in GSCs.²³⁰ They observed higher levels of NEAT1 in GSCs, as well as in the serum of GBM patients. Suppression of NEAT1 was able to prevent GSCs from proliferating, migrating, and invading. Similar results were found after upregulation of let-7g-5p, which was identified as a direct target of NEAT1. Next, they found that let-7g-5p exerted its effects by targeting and inhibiting MAP3K1. Taken together, their data suggested that NEAT1 could promote GSC pro-oncogenic activity and TMZ tolerance by regulating the let-7g-5p/MAP3K1 pathway.²³⁰

Esophageal squamous cell cancer and glioma have both been reported to show increased levels of MALAT1.^231,232 miR-129-3p, miR-129-2-3p, and miR-129–5p are three important members of the miR-129 family.²³³ miR-129–5p has tumor-suppressor roles in various cancers, such as ovarian,²³⁴ breast,²³⁵ and glioma.²³⁶ In glioma, miR-129 overexpression showed an anti-oncogenic effect by regulating the Notch-1/E2F7/Beclin-1 pathway.²³⁷ SOX2 is regarded as a molecular signature of GSCs, as well as pluripotent stem cells.^238,239 Xiong et al. examined the role of MALAT1 in GSCs and the interactions between this lncRNA with miR-129 and SOX2.²⁴⁰ Compared with regular (non-stem) glioma cells, GSCs showed higher levels of MALAT1, but lower levels of miR-129. MALAT1 inhibition could impair GSC proliferation by upregulating miR-129. miR-129 was found to target and inhibit SOX2. These effects were also observed in vivo. They concluded that MALAT1 could promote GSC tumorigenicity both in vitro and in vivo, by regulating the miR-129/ SOX2 axis.²⁴⁰

lncRNA TP73-AS1 was found to be epigenetically downregulated in both oligodendroglioma and GBM.^241,242 In addition, GBM patient prognosis was found to be positively correlated with TP73-AS1 levels.²⁴³ ALDH1A1 has been identified as a marker of GSCs,^244,245 and is involved in GSC progression and therapy resistance.245, 246, 247 It was proposed that ALDH1A1 could interfere with the oxidative stress triggered by chemotherapeutic drugs and induce tolerance to treatment.²⁴⁶ Mazor et al. studied the effects of TP73-AS1 in GBM patients and GSCs.²⁴⁸ High levels of TP73-AS1 were observed in GBM patients and were correlated with poor clinical outcomes. They found that TP73-AS1 could attenuate the response of GSCs to TMZ treatment, which could be attributed to its ability to regulate ALDH1A1. Finally, they found a correlation between TP73-AS1 overexpression and poor clinical outcomes in GBM patients. They suggested that TP73-AS1 could enhance tumorigenicity in GSCs and reduce their sensitivity to TMZ by upregulating ALDH1A1.²⁴⁸

XIST (X-inactive-specific transcript) is a lncRNA gene on the X chromosome of placental mammals, which produces a lncRNA to silence one of the paired X chromosomes in females.²⁴⁹ Aberrant XIST expression has been observed in several cancers, and its oncogenic effect may be explained by causing instability in the heterochromatin structure.²⁵⁰ Furthermore, lncRNA XIST may promote the viability of hematopoietic stem cells.²⁵¹ Yao et al. investigated the role of XIST in human GSCs.²⁵⁰ They found that both glioma cells and GSCs had elevated levels of XIST. In vitro downregulation of XIST in GSCs reduced proliferation, migration, and invasion, while promoting apoptosis and suppressing oncogenesis. The same results were observed after XIST downregulation in a murine model. They identified miR-152 as a direct target of XIST to explain its function. miR-152 has been shown to exert anti-oncogenic effects in GSCs by regulating KLF4.³⁷ In conclusion, they identified XIST as an oncogene whose suppression could reduce the oncogenesis of GSCs by upregulating miR-152.²⁵²

Fibroblasts are a group of stromal cells in the tumor microenvironment (TME), which may promote tumor cell progression and metastasis.^253,254 Cancer cells can progressively activate normal fibroblasts within their environment to form cancer-associated fibroblasts (CAFs).²⁵⁵ Because of the importance attributed to CAFs in the TME, investigating their potential as targets to treat gliomas is of increasing interest.²⁵⁶ There is an association between the CAF abundance in the TME and poor clinical outcomes.^257,258 lncRNA HOTAIRM1 shares the same location as HOX genes and has been reported to have pro-oncogenic or anti-oncogenic effects in various cancers by regulating HOXA genes.259, 260, 261, 262, 263 The anti-oncogenic effects of miR-133b have been proposed to be mediated by different molecules in different cancers, including HOTAIRM1.264, 265, 266, 267, 268 Wang et al. explored the interaction between GSCs and fibroblasts in TME, both in vitro and in vivo, and the underlying molecular pathways.²⁶⁹ GSCs were able to trigger fibroblasts to behave as malignantly transformed fibroblasts (t-FBs). They observed elevated levels of HOTAIRM1 in both glioma cells and t-FBs. In addition, a correlation between high HOTAIRM1 levels and poor clinical outcomes was observed in glioma patients. HOTAIRM1 knockdown suppressed the pro-tumorigenic and malignant behavior of t-FBs, while the opposite effect was observed by HOTAIRM1 upregulation. At the molecular level, HOTAIRM1 was found to target and inhibit miR-133b-3p, which in turn upregulated TGF-β. Taken together, HOTAIRM1 was identified as an oncogene in GSCs, which could modulate the t-FB malignant behavior by inhibiting miR-133b-3p and upregulating TGF-β.²⁶⁹

Table 2 lists some lncRNAs that have been reported to be involved in CSCs, and GSCs in particular.

Table 2.

Role of lncRNAs in cancer stem cells

LncRNA	Expression	Target	Model (in vitro, in vivo, human)	Ref.
TUG1	↑	Nestin, miR-145, SOX2, MYC, PRC2 components (EZH2, SUZ12), YY1, BDNF, NGF, NTF3	in vitro, in vivo	Katsushima et al.270
LINC00115	↑	miR-200s, ZEB1, ZNF596/EZH2/STAT3 signaling pathway	in vitro	Tang et al.271
MALAT1	↑	miR-129-5p, HMGB1	in vitro	Yang et al.272
Linc00152	↑	miR-103a-3p/FEZF1/CDC25A axis	in vitro, in vivo	Yu et al.273
GAS5	↓	miR-196a-5p/FOXO1/PID1, MIIP pathway	in vitro, in vivo	Zhao et al.274
HOTAIRM1	↑	HOX genes	in vitro, in vivo	Xia et al.275
PCAT1	↑	miR-129–5p, HMGB1	in vitro	Zhang et al.276
NEAT1	↑	let-7g-5p/MAP3K1	in vitro, human	Bi et al.230
MALAT1	↑	miR-129, SOX2	in vitro, in vivo	Xiong et al.240
NEAT1	↑	let-7e, NRAS	in vitro, in vivo	Gong et al.222
TP73-AS1	↑	ALDH1A1	in vitro, human	Mazor et al.248
CRNDE	↑	miR-186/XIAP, PAK7	in vitro, in vivo	Zheng et al.277
TALNEC2	↑	miR-21, miR-191	in vitro, in vivo	Brodie et al.278
NEAT1	↑	miR-107, CDK6	in vitro	Yang et al.279
XIST	↑	miR-152	in vitro, in vivo	Yao et al.252
SNHG9	↑	miR-326/SOX9	in vitro	Wang et al.280
Linc01060	↑	MZF1/c-Myc/HIF1α	in vitro, in vivo, human	Li et al.281
MIR22HG	↑	miR-22-3p, miR-22-5p, SFRP2, PCDH15, Wnt/β-catenin	in vitro, in vivo, human	Han et al.282
ASB16-AS1	↑	E-cadherin, N-cadherin, vimentin, EMT	in vitro, human	Zhang et al.283
lncRNA-ZNF281	↓	NF-κB1	in vitro, in vivo	Li et al.284
MALAT1	↑	MRP1, Bcl-2, HSP70, IAPs, p53	in vitro, in vivo	Kim et al.285
ENSG00000235427.1	↓	CAV1	in vitro	Li et al.286
ENSG00000261924.1	↓	RPTOR	in vitro	Li et al.286
P2RX5-TAX1BP3	↓	TAX1BP3	in vitro	Li et al.286
MALAT1	↑	ERK/MAPK	in vitro	Han et al.287
lincRNA-ROR	↓	KLF4	in vitro	Feng et al.288
H19	↑	–	in vitro	Li et al.289
HIF1A-AS2	↑	IGF2BP2, DHX9, HMGA1	in vitro, in vivo	Mineo et al.290
TUG1	↓	EZH2	in vitro, in vivo	Cao et al.291
HOXB-AS1	↑	–	in vitro	Shao et al.292
H19	↑	–	in vitro, in vivo, human	Jiang et al.293
SOX2OT	↑	miR-194-5p, miR-122, SOX3, TDGF-1	in vitro, in vivo	Su et al.294
RP11-279C4.1	↑	miR-1273g-3p/CBX3	in vitro, in vivo	Wang et al.295
HOTAIR	↑	EZH2, LSD1, PDCD4, CCND1, CDK4	in vitro, in vivo	Fang et al.296
MEG3	↓	vimentin, β-actin, Src_pY527, FAK_pY397, caveolin-1, connexin-43, NDRG1_pT346	in vitro, in vivo, human	Buccarelli et al.297
HOTAIRM1	↑	miR-133b-3p/TGF-β	in vitro, in vivo, human	Wang et al.269

Enhancer RNAs

Enhancer RNAs (eRNAs), a new subclass of lncRNAs, participate in the regulation process of gene transcription.²⁹⁸ A growing number of studies showed that eRNAs interact with transcription factors, RNA-binding proteins, and transcriptional coactivators, such as CBP/p300 and Bromodomain-containing protein 4.²⁹⁹ Another mechanism discovered to underlie eRNA functions is that eRNAs participate in transcription factor trapping to increase their local concentration at DNA at the site of transcription.^299,300

Based on the evidence, eRNAs play a critical role not only in cell development and homeostasis but also indirectly drive human diseases and differentiation. The most recent findings provide new insights into the characteristics and mechanisms of action of eRNAs, highlighting potentially broad roles of eRNA interactions in tumorigenesis and various cancer types.^301,302 By modifying gene transcription and protein-RNA interactions, they can influence the expression of oncogenes and tumor-suppressor genes, as well as in abnormal cellular responses to external signals, such as inflammation, hypoxia, hormones, and other stimuli.^300,303 Emerging studies also indicated the role of eRNAs in the regulation of key immune checkpoints and immune escape of tumor cells.^301,302 For example, CCAT1, a super-enhancer-derived eRNA, induces PD-L1 expression via activating PI3K/AKT and RAS/MAPK pathways.³⁰⁰

Many eRNAs were found to be significantly overexpressed in tumor samples when compared with adjacent normal tissues. Because of their cancer-specific pattern of expression, eRNAs are clinically relevant and can serve as diagnostic, prognostic, and treatment response biomarkers in cancer therapy.^304,305 In this setting, thanks to the efforts of scientists who attempted to infer cancer-specific expression of eRNAs from RNA sequencing data collected in numerous cancer series around the world,^306,307 a systematic mapping of eRNAs expressed in various types of cancer is now available. The expression profiles of those eRNAs may help in eliminating intratumor heterogeneity and improving the diagnosis and treatment of a variety of cancers. For instance, focally amplified lncRNA on chromosome 1 (FAL1) has been recognized as an oncogene in numerous cancers and its overexpression is usually associated with poor prognosis.³⁰⁸ It supports cell proliferation and facilitates EMT, migration, and invasion by modulating the PTEN/AKT pathway. In addition, FAL1 contributes to the growth and metastatic potential of cancer cells via STAT3 phosphorylation and phosphorylation of GSK-3β, a protein crucial in Wnt signaling pathway regulation.³⁰⁸

Inflammatory signals have been shown to activate extensive programs of enhancer activation and eRNA production. Rahnamoun et al. revealed, in cancer cells, that p53 mutants abnormally activated a group of enhancers in response to pro-inflammatory TNF-α signaling.³⁰⁹ Co-binding of mutant p53 and NF-κB at these enhancers induced eRNA synthesis, one of which was necessary for the activation of key inflammation genes, such as C-C motif chemokine ligand 2 (CCL2).³¹⁰ As a result, eRNAs play a direct role in cancer cell immune response. Many other cancer-related signaling pathways, including the Wnt, Notch, and Hippo pathways, orchestrate nuclear events, such as chromatin remodeling and transcription factor/cofactor recruitment to function by enhancer control.³⁰³

Lin et al. recently used the PreSTIGE computational pipeline to predict tissue-specific enhancer-derived RNAs and the underlying regulatory genes.³¹¹ They chose three eRNAs for their significant prognostic values to construct a risk signature: CRNDE, LINC00844, and MRPS31P5. Pathway and gene ontology analyses revealed that the risk signature in glioma is associated with mRNA processing and spliceosome. Furthermore, they discovered that hub eRNAs may regulate the expression of a variety of splicing factors, including MOV10 and SEC31B, and are associated with prognosis-associated alteration splicing. The researchers developed a risk signature composed of three eRNAs that can be used as targets to accurately predict prognosis in glioma patients.³¹¹ In another study, Guo et al., by functional enrichment analysis and immunogenomic profiling, indicated that AC003092.1 as an immune-related eRNA is related to glioma-immunosuppressive microenvironment.³¹²

circRNAs and CSCs in glioma

The circRNA Serpine2 is able to regulate the migration and invasion of glioma cells by modulating the expression of uPA and MMP-9/2.³¹³ It was also able to promote the transformation of preneoplastic lesions into medulloblastoma.³¹⁴ The tumor-suppressor role of miR-124-3p has been reported in glioma and GSCs.^175,315 KIF20A was found to be a hub gene with an important function in p53 regulation,³¹⁶ and was also highly elevated during mitotic processes in glioma cells.³¹⁷ Using in silico prediction, Li and Lan tried to identify circRNA/miRNA/mRNA axes in GSCs,³¹⁸ resulting in the identification of the Serpine2/miR-124-3p/KIF20A axis. Serpine2 was found to be upregulated in GSCs and could be delivered to glioma cells inside exosomes, to enhance glioma cell tumorigenesis. Serpine2 exerted its effects by inhibiting miR-124-3p and upregulating KIF20A. All these results were replicated in vivo. Serpine2 was identified as an oncogenic agent in GSCs, which could be transported to glioma cells to enhance their tumorigenicity by regulating the miR-124-3p/KIF20A axis.³¹⁸

EGFR is reported to be highly expressed in about 50% of GBM tumors and has been recognized as an oncogene in GBM.^319,320 Many studies have attempted to target EGFR to treat GBM, but the results have not so far been very successful.321, 322, 323 It has been found that circRNAs are readily translated³²⁴ because they lack a stop codon in their structure.³²⁵ Gao et al. investigated EGFR activity in GBM³²⁶ and discovered an additional pathway for activating EGFR independent of EGF. In this pathway, C-E-Cad (a variant of E-cadherin) was found to act as a ligand for EGFR. C-E-Cad is translated from circ-E-Cad, a translatable circRNA with high expression levels in GSCs, which enhances their tumorigenicity. Moreover, the efficacy of anti-EGFR therapy was significantly increased by suppressing C-E-Cad expression. In conclusion, they identified C-E-Cad as an independent activating ligand for EGFR in GBM, which could be targeted to improve the efficacy of anti-EGFR therapy.³²⁶

The Hedgehog (HH) signaling pathway is activated in various cancers and plays an important role in embryonic stem cells while it is silent in mature cells.327, 328, 329 The HH network includes HH ligands (Shh, Ihh, and Dhh), as well as PTCH, SMO, and Gli proteins.³³⁰ The HH pathway works as follows: first HH binds to PTCH to derepress SMO, then SMO prevents SUFU from inhibiting Gli1, and then the activated transcription factor Gli1 can regulate gene expression. Direct inhibition of SMO via PTCH has not yet been proven, but it has been found that cholesterol is needed to prevent PTCH from inhibiting SMO. In addition, cholesterol can endogenously activate SMO.³³¹ Nevertheless, the exact mechanism for PTCH suppression of SMO is elusive, and understanding this step could clarify the whole HH pathway.³³² Wu et al. explored the details of the HH pathway in GBM.³³³ They discovered a new protein called SMO-193a.a, which affects the HH pathway. SMO-193a.a is translated from circ-SMO (a translatable circRNA) in GSCs. Knockdown of SMO-193a.a disrupted the HH pathway in GSCs and reduced their tumorigenic ability both in vitro and in vivo. In addition, Gli1 could target FUS to upregulate SMO-193a.a, and the HH pathway activity is maintained in GSCs via the Shh/Gli1/FUS/SMO-193a.a axis. Clinically speaking, SMO-193a.a protein expression is more specific for GBM than SMO RNA expression and is better correlated with Gli1 levels. Furthermore, they also observed a correlation between SMO-193a.a levels and a poor prognosis in GBM patients. They concluded that SMO-193a.a could be translated from circSMO to increase the oncogenic capacity of GSCs via induction of the HH pathway.³³³

Recently, significant overexpression of circSCAF11 was discovered in glioma tissues and cell lines, and ectopic upregulation of circSCAF11 was found to be closely related to glioma patients' poor clinical outcome.³³⁴

Table 3 lists some circRNAs that have been reported to be involved in CSCs, and GSCs in particular.

Table 3.

Role of circRNAs in cancer stem cells

Circular RNAs	Expression	Target	Model (in vitro, in vivo, human)	Ref.
circPTN	↑	miR-145-5p/miR-330–5p	in vitro, in vivo	Chen et al.10
Serpine2	↑	miR-124-3p/KIF20A	in vitro, in vivo	Li and Lan318
circCHAF1A	↑	FMR1/circCHAF1A/miR-211-5p/HOXC8, MDM2, p53	in vitro, in vivo, human	Jiang et al.335
cMELK	↑	miR-593/EphB2	in vitro, in vivo	Zhou et al.336
circATP5B	↑	miR-185-5p/HOXB5, JAK2/STAT3	in vitro, in vivo, human	Zhao et al.337
circ-E-Cad (translatable)	↑	EGFR-STAT3	in vitro, in vivo, human	Gao et al.326
circ-SMO (translatable)	↑	SMO	in vitro, in vivo	Wu et al.333
cARF1	↑	miR-342-3p/ISL2	in vitro, in vivo, human	Jiang et al.338

Conclusions

The properties of stem cells are self-regeneration and differentiation into several lineages of normal cells, but CSCs may be caused by disturbance of these properties. The presence of CSCs in a tumor causes metastasis to spread more readily. In the brain, the rate and developmental timing of neurogenesis can be changed by the differentiation and self-renewal of cortical progenitor cells. The main reason for the development of glioma is a failure of cellular differentiation, but there is also evidence that aberrant epigenetic mechanisms involving ncRNAs are involved in glioma development. NcRNAs can regulate cellular signaling in CSCs and glioma cells. However, more research into the exact pathways and mechanisms of action of ncRNAs in CSCs and glioma is required to develop a more effective therapy for glioma patients. Recently, new studies have revealed the role of lncRNAs in embryonic pluripotency and self-renewal potential, but there is still a need for more studies into the exact role of ncRNAs in the transformation process, CSC therapy resistance, and maintaining stemness. Thus, ncRNAs could allow us to eventually achieve more success in glioma treatment. These in-depth studies of ncRNA biology will ultimately yield further insight into the molecular mechanisms of tumorigenesis, and lead to the development of improved therapeutic strategies against glioma, which are urgently needed. A summary of the function of glioma stem cell ncRNAs and its mechanism is given in Table 4.

Table 4.

The summary of tumor stem cells ncRNAs role in glioma

ncRNA	Type	Effect	Mechanism	Ref.
FOXD2-AS1	lncRNA	promoting stemness and proliferation	recruiting TAF-1 to the NOTCH1 promoter region	Wang et al.339
circ-Serpine2	lncRNA	promoting proliferation, migration, and invasion	circ-Serpine2 could upregulate KIF20A by sponging miR-124-3p	Li and Lan318
RBM5-AS1	lncRNA	promotes radioresistance in medulloblastoma	stabilization of SIRT6 protein	Zhu et al.340
TUG1	lncRNA	alleviated TMZ resistance and inhibited tumorigenicity	downregulating EZH2 expression	Cao et al.291
SNHG9	lncRNA	facilitates growth of GSCs	competitive endogenous RNA of miR-326 to elevate the expression of SOX9	Wang et al.341
RP11-279C4.1	lncRNA	functions as an oncogene that promotes tumour progression	modulating the miR-1273g-3p/CBX3 axis	Wang et al.295
TPTEP1	lncRNA	inhibits stemness and radioresistance	miR-106a-5p-mediated P38 MAPK signaling	Tang et al.342
LINC01057	lncRNA	promotes mesenchymal differentiation	activating NF-κB	Tang et al.343
NEAT1	lncRNA	promotes malignant phenotypes and TMZ resistance in GBM stem cells	MAP3K1, as a direct target of let-7g-5p, is positively regulated by NEAT1	Bi et al.230
SNHG20	lncRNA	promotes tumorigenesis and cancer stemness	activating PI3K/Akt/mTOR signaling pathway	Gao et al.344
TP73-AS1	lncRNA	promotes TMZ resistance	regulation of the expression of metabolism-related genes and ALDH1A1	Mazor et al.248
PCAT1	lncRNA	PCAT1 knockdown restrained the sphere-formation ability, increased the apoptosis rate and DNA damage under radiation treatment	increase the expression of miR-129-5p and decrease the expression of HMGB1	Zhang et al.276
MALAT1	lncRNA	siRNA against MALAT1 sensitizes GBM to TMZ	–	Kim et al.285
SOX2OT	lncRNA	knockdown of SOX2OT inhibits the malignant biological behaviors	upregulating the expression of miR-194-5p and miR-122	Su et al.294
TALNEC2	miRNA	increased tumorigenic potential of GSCs and their resistance to radiation	downregulation of miR-21 and miR-191	Gao et al. and Brodie et al.141,278
miR-103a	miRNA	decreased the radioresistance capability	suppressing the FGF2-XRCC3 axis	Gu et al.345
miR-139	miRNA	inhibitory functions on GSC stemness and tumorigenesis	inhibiting Wnt/β-catenin signalling	Li et al.202
miR-27a-5p	miRNA	enhanced the sensitivity of glioma stem cells to radiotherapy	shFOSL1-inhibited miR-27a-5p expression	Li et al.346
miR-944	miRNA	reduces glioma growth and angiogenesis	inhibiting AKT/ERK signalling	Jiang et al.216
miR-128, miR-302a	miRNA	enhances senescence-associated cytotoxicity of axitinib to overcome drug resistance	–	Cardoso et al.347
miR-30b-3p	miRNA	confer TMZ resistance	directly targeting RHOB	Yin et al.156
miR-146b-5p	miRNA	suppresses the malignant phenotype	miR-146b-5p inhibited SMARCA5 expression and inactivated a TGF-β pathway	Wang et al.127
mir-370-3p	miRNA	inhibiting glioma cell growth, migration, and invasion	targeting the NEAT1, HMGA2, and HIF1A	Lulli et al.140
miR-603	miRNA	simultaneously promoted the CSC state and upregulated DNA repair to promote acquired resistance	targeting IGF1 and IGF1R	Ramakrishnan et al.136
miR-27a-3p, miR-22-3p, miR-221-3p	miRNA	exacerbated radiotherapy resistance	targeting CHD7	Zhang et al.348
miR-486-5p	miRNA	enhanced the self-renewal capacity	miR-486-5p as a Sox2-induced miRNA that targets the tumor-suppressor genes PTEN and FoxO1	Lopez-Bertoni et al.134
miR-30a	miRNA	suppresses self-renewal and tumorigenicity	blocking the NT5E-dependent Akt signaling pathway by targeting the NT5E	Peng et al.124
miR-124	miRNA	promotes a stem-like to neuronal transition, with reduced tumorigenicity and increased radiation sensitivity	targeting the SOX9 and inhibition of ERK1/2	Sabelström et al.349
miR-181d	miRNA	interferes in the GBM CSC response to treatment with TMZ and ionizing radiation	miR-181d associated with the methylation status of the MGMT	Lizarte Neto et al.350
miR-93	miRNA	enhanced the activity of IR and TMZ against GSCs	simultaneous inhibition of multiple autophagy regulators, including BECN1/Beclin 1, ATG5, ATG4B, and SQSTM1/p62	Huang et al.121
miR-7-5p	miRNA	suppresses stemness and enhances TMZ sensitivity of drug-resistant GBM	targeting Yin Yang 1 (YY1)	Jia et al.142
miR-186	miRNA	reverses cisplatin resistance and inhibits the formation of the GBM	degrading Yin Yang 1	Li et al.351
miR-29a	miRNA	improved sensitivity to cisplatin	–	Yang et al.352
miR-132	miRNA	induces TMZ resistance and promotes the formation of CSC phenotypes	targeting TUSC3	Cheng et al.353
miR-223	miRNA	increase the sensitivity of glioma to TMZ	regulating PI3K/Akt signaling pathway	Huang et al.165
let-7g-5p	miRNA	inhibits epithelial-mesenchymal transition consistent with reduction of glioma stem cell phenotypes	targeting VSIG4	Zhang et al.225
miR-146b-5p	miRNA	attenuates stemness and radioresistance	targeting HuR/lincRNA-p21/β-catenin pathway	Yang et al.153
miR-218-5p	miRNA	inhibits the stem cell properties and invasive ability	reduced stem cell marker (A2B5, nestin, PLAGL2, ALDH1 and Sox2) expression	Wu et al.196
miR-125b	miRNA	sensitize TMZ-induced anti-glioma stem cancer effects	inactivation of Wnt/β-catenin signaling pathway	Shi et al.354
miR-153	miRNA	decreased radioresistance and stemness	targeting Nrf-2/GPx1/ROS pathway	Yang et al.166
miR-30	miRNA	promotes glioma stem cells	decreased the expression of suppressor of cytokine signaling 3 (SOCS3) expression	Che et al.148
miR-210	miRNA	miR-210 knockdown decreases hypoxic glioma stem cells stemness and radioresistance	–	Yang et al.355
miR-455-3p	miRNA	TMZ resistance	–	Tezcan et al.176
miR-125b	miRNA	enhance the chemosensitivity of GBM stem cells to TMZ	targeting Bak1	Chen et al.177
miR-125b	miRNA	inhibition of miR-125b enhance sensitivity of GSCs to TMZ	targeting PIAS3	Shi et al.209
miR-17	miRNA	decreased cell proliferation and drug resistance	repress MDM2	Li and Yang356
miR-23b	miRNA	enhanced the sensitivity to TMZ	–	Geng et al.172
miR-145	miRNA	reduced chemoradioresistance	targeting Oct4 and Sox2	Yang et al.357
miR-125b-2	miRNA	resistance to TMZ	mitochondrial pathway of apoptosis	Chan et al.158
miR-9	miRNA	suppression of miR-9 confer stemness potential and chemoresistance	induces SOX2	Jeon et al.358
miR-328	miRNA	decrease the chemoresistance	targeting ABCG2	Li et al.359

Availability of data and material

The primary data for this study are available from the authors on request.