Micro-masters of glioblastoma biology and therapy: increasingly recognized roles for microRNAs

Abstract

MicroRNAs are small noncoding RNAs encoded in eukaryotic genomes that have been found to play critical roles in most biological processes, including cancer. This is true for glioblastoma, the most common and lethal primary brain tumor, for which microRNAs have been shown to strongly influence cell viability, stem cell characteristics, invasiveness, angiogenesis, metabolism, and immune evasion. Developing microRNAs as prognostic markers or as therapeutic agents is showing increasing promise and has potential to reach the clinic in the next several years. This succinct review summarizes current progress and future directions in this exciting and steadily expanding field.

MicroRNAs are small noncoding RNAs discovered recently in humans but with an ever-increasing impact across the biological sciences. This is true for oncology, particularly for brain tumors such as glioblastoma. MicroRNAs play oncogenic, tumor-suppressive, and prognostic roles in glioblastoma and may provide novel therapies. This review provides a succinct update on the current state of the art of microRNAs in glioblastoma, giving context for 2 valuable additions to the field described in this issue.

Encoded in our genome are 1500–2000 microRNAs. They originate as small hairpins within larger “primary microRNA” transcripts from introns or intergenic regions, either as individual microRNAs or as clusters on the same transcript. The microRNAs arise from 60–80 base-pair segments with 2 partially complementary regions that fold into a hairpin. These hairpins are recognized and cleaved out from the primary microRNA by the DGCR8 (DiGeorge syndrome critical region 8) protein complex, releasing pre-microRNAs that exit the nucleus via transporters such as exportin-5.¹ In the cytoplasm, the loop region on the pre-microRNA hairpins is cleaved off by the Dicer complex, yielding 18–22 bp double-stranded mature microRNAs. Mature microRNAs are recognized by the RNA-induced silencing complex, which cleaves one microRNA strand but uses the other to identify partially complementary sites in the 3′-untranslated region (UTR) or sometimes the 5′-UTR of target mRNAs (microRNA processing/function illustrated in Fig. 1).² Generally there must be “seed matches” with perfect complementarity at bases 1–7 or 2–8 of the microRNA. This triggers suppressed translation or degradation of target mRNAs. More seed matches or greater microRNA/target complementarity leads to stronger target suppression, but criteria such as additional complementarity and location of target sites in the 3′-UTR are also important.³ Notably, each microRNA has many targets, and most genes are targeted by multiple microRNAs. There is a common misconception that microRNAs lead to modest target suppression, but the best targets for a given microRNA are actually quite powerfully regulated—with effects comparable to efficient short interfering RNAs.

Fig. 1.

MicroRNA expression, processing, and function. RISC, RNA-induced silencing complex; Ran-GTP, Ras-related nuclear protein–guanosine triphosphate; TAR, HIV transactivation response. 

MicroRNAs are linchpins in innumerable biological processes, including cancer. A number of microRNAs were shown to play tumor-suppressive roles in cancer by reducing expression of oncogenes, while other microRNAs were found to have oncogenic roles (“oncomiRs”) by suppressing expression of tumor suppressor genes. Early and important discoveries included the oncomiRs miR-21 and the miR-17–92 cluster, as well as the let-7 family of tumor-suppressive microRNAs.^4–6 Given the prominence of microRNAs in cancer and the rich expression of brain-specific microRNAs, it is not surprising that a steadily increasing number of reports have shown important roles for microRNAs in glioblastoma, the most common and lethal primary brain cancer (annual citations plotted in Fig. 2). MicroRNAs play important roles in multiple facets of glioblastoma (due to space constraints, only a limited number will be mentioned here, summarized in Table 1—we regret the necessity of omitting many valuable studies).

Fig. 2.

PubMed citations with search terms “microRNA” and “glioblastoma” by year.

Table 1.

Selected microRNAs shown to play a role in glioblastoma multiforme (GBM)

MiRNA	Type	References	Targets or Regulators	Physiological Significance in GBM
MiR-21	OG	4, 7–9, 45	PDCD4, PTEN, tropomyosin 1(α)	CV, INV, BM
MiR-17–92	OG	5, 17, 19	Nemo-like kinase	CV, INV
Let-7	TS	6, 27	Ras, HMGA2, blocked by LIN28A	CV, INV
MiR-10b	OG	21, 41	TGF-β	CV, BM
MiR-34a	TS	12, 20	Notch, c-Met, TGF-β	CV, SC, proneural
MiR-7	TS	11	EGFR	CV, SC
MiR-124-3p	TS	13, 33	STAT3, FoxP3	CV, INV, SC, IE
MiR-124-5p	TS	Current issue	Laminin B1	EX-LT
MiR-137	TS	13, 25	RTVP-1	CV, SC
MiR-326	TS	10, 31, 43	Notch	CV, SC, MB, BM
			PKM2
MiR-99a	TS	14	FGFR3	CV, SC
MiR-524-5p	TS	15	Jagged-1, HES1	SC
MiR-328	OG	18	SFRP1	SC, INV
MiR-128	TS	22	BMI-1	SC
MiR-101	TS	23	EZH2	SC
MiR-302–367	TS	24	CXCR4	SC
MiR-143	OG	26		INV
MiR-145	OG	26		INV
MiR-218	OG	28	RTK-HIF	SC, AG, mesenchymal
MiR-93	OG	29	Integrin-β8	INV, AG
MiR-125b	TS	30	MAZ	AG
MiR-451	TS	32	Liver kinase B1 binding partner	CV, MB
MiR-222	OG	34	ICAM-1	CV, IE
MiR-339	OG	34	ICAM-1	IE
MiR-148a	TS	42	MIG6, Bim	CV, SC, BM
MiR-181d	TS	44	MGMT	BM
MiR-297	TS	48, 50	DGKα, VEGFA	AG, new targets
				EX-LT
MiR-196a	OG	Current issue	IκBα	EX-LT for antagomir

Abbreviations: Type = OS, oncogenic; TS, tumor-suppressive. Targets/Regulators = BMI-1, B lymphoma Moloney murine leukemia virus insertion region 1 homolog; CXCR4, C-X-C chemokine receptor type 4; DGKα, diacylglycerol kinase alpha; EGFR, epidermal growth factor receptor; EZH2, enhancer of zeste homolog 2; FGFR3, fibroblast growth factor receptor 3; FoxP3, forkhead box P3; HES1, hairy and enhancer of split–1; HMGA2, high-mobility group adenine thymine–hook 2; ICAM-1, intercellular adhesion molecule 1; IκBα, inhibitor of NF-κB alpha; MGMT, O⁶-DNA methylguanine-methyltransferase; MIG6 = mitogen-inducible gene 6; PDCD4, programmed cell death 4; PKM2, pyruvate kinase M2 isoform; PTEN, phosphatase and tensin homolog; RTK-HIF, receptor tyrosine kinase–hypoxia-inducible factor; RTVP-1, related to testes-specific, vespid, and pathogenesis proteins 1; SFRP1, secreted frizzled-related protein 1; STAT3, signal transducer and activator of transcription 3; VEGFA, vascular endothelial growth factor A. Physiological Significance = CV, cell viability; INV, invasiveness; MB, metabolism; SC, stem cells; subtype association (listed with entire name of subtype, such as “proneural”) = AG, angiogenesis; BM, biomarker; EX-LT, established xenograft-local treatment model; IE, immune evasion.

Glioblastoma Cell Viability

A number of potent oncogenic microRNAs have been described for glioblastoma that promote cell viability by inhibiting expression of tumor suppressors. MicroRNA-21 is the first prominent example, and perhaps the most widely studied oncomiR. Key targets include programmed cell death 4, phosphatase and tensin homolog, tropomyosin 1(α), and others.^6–8 Efforts are under way to deliver miR-21 inhibitors as therapy for glioblastoma, potentially combined with other treatments. MiR-10b is another powerful oncomiR, and has been shown to influence glioblastoma cell survival and patient prognosis in a report that included analysis of The Cancer Genome Atlas (TCGA) data and a local treatment model using anti–miR-10b.⁹ In contradistinction to these oncomiRs, numerous tumor-suppressive microRNAs have been identified that are underexpressed and cytotoxic in glioblastoma. Examples include miR-34a, miR-7, miR-124, miR-137, and miR-326, which inhibit expression of numerous critical oncogenes/oncogenic pathways such as Notch, c-Met, epidermal growth factor receptor, and phosphatidylinositol-3 kinase/Akt.^10–13 Interestingly, while many of the oncomiRs identified to date are shared across many cancers, the tumor-suppressive microRNAs are often restricted to glioblastoma. Besides reducing expression of tumor-suppressive microRNAs, some glioblastomas may incorporate other “escape routes” for microRNA-targeted oncogenic proteins; a recent study found that one function of a fibroblast growth factor receptor 3/transforming acidic coiled-coil 3 fusion protein was to eliminate the fibroblast growth factor receptor 3 3′-UTR and avoid regulation by miR-99a.¹⁴

Glioblastoma Stemlike Cells/Tumor-initiating Cells and Stem Cell Pathways

In recent years it has become increasingly well established that glioblastomas and many other cancers harbor a subpopulation of stemlike cells that are highly tumorigenic and possibly resistant to standard therapies. MicroRNAs may have especially important roles in these glioblastoma stemlike cells and in addressing the therapeutic challenge they present, in part because microRNAs strongly regulate stem cell pathways/proteins such as Notch, Hedgehog, Wnt, transforming growth factor (TGF)–β, and polycomb complexes. MiR-34a, miR-326, miR-524-5p, and others have all been reported to target Notch family members and ligands, and miR-326 has also been shown to suppress the Hedgehog pathway.^10,12,15,16 The Wnt pathway has been found to be regulated in glioblastoma both by the oncomiR miR-92b, which targets the Wnt-inhibiting NLK gene, and by miR-328 suppression of SFRP1 as well.^17,18 TGF-β signaling and its collaborator nuclear factor–kappaB (NF-κB) have been identified as important in glioblastoma stem cells, and reports have shown these mesenchymal/inflammatory pathways to be regulated in glioblastoma by miR-34a, miR-10b, and miR-17–92.^19–21 The polycomb complexes repress gene expression to promote stem cell function; their components Bmi-1 and enhancer of zeste homolog 2 are critical in glioblastoma stem cells and are regulated by miR-128 and miR-101.^22,23 On another front, differentiation of glioblastoma stem cells was shown to increase expression of the tumor-suppressive miR-302–367 cluster, which was found to control stem cell function through regulation of CXCR4, which in turn modulated expression of the stem cell regulators SHH, GLI1, and NANOG.²⁴ A second report has shown that the tumor-suppressive miR-137 targets RTVP-1 (related to testes-specific, vespid, and pathogenesis proteins 1) to indirectly regulate CXCR4 by another pathway.²⁵

Glioblastoma Migration/Invasion

Much of the lethality of glioblastoma is due to relentless brain invasion, driven in part by microRNAs. The oncomiR miR-21 promotes glioblastoma invasiveness through suppressing expression of matrix metalloprotease inhibitors, and miR-328 also increases invasiveness via regulation of the Wnt pathway.^9,18 Glioblastoma cells selected for higher invasiveness demonstrated upregulated miR-143 and miR-145 expression, and these microRNAs were then shown to mediate this invasive phenotype.²⁶ MicroRNAs have also been identified as key members of circuits driving glioblastoma invasiveness. The let-7 family of tumor-suppressive microRNAs is inhibited by Lin28A, which is normally expressed in development but is overexpressed in glioblastoma, as indicated by TCGA data. There is a strong correlation in glioblastoma between Lin28A expression and expression of the pro-invasive HMGA2 gene targeted by let-7 microRNAs, and an anticorrelation with let-7 family members.²⁷ This report also showed that overexpression of let-7g can reverse the invasive phenotype of Lin28A-expressing glioblastoma stem cells.

Angiogenesis

New blood vessel growth feeds the growth of glioblastomas, and here too microRNAs loom large. MiR-218 was recently shown in glioblastoma to regulate a circuit with receptor tyrosine kinases and the hypoxia-inducible factors that control angiogenesis.²⁸ MiR-93 plays a role in glioblastoma-associated angiogenesis by targeting integrin B8, a tumor suppressor and inhibitor of angiogenesis.²⁹ MiR-93 was sufficient to enhance angiogenesis and tumor growth and drastically reduce survival in a subcutaneous xenograft model of glioblastoma. MiR-125b has been shown to be downregulated in both human glioblastoma-associated endothelium and in endothelial cells cultured with conditioned medium from glioblastoma cells.³⁰ This work identified Myc-associated zinc finger protein (MAZ) as a target of miR-125b and demonstrated that MAZ expression is increased in glioblastoma-associated endothelium and is driven by vascular endothelial growth factor.

Metabolism

Aberrant metabolism is a hallmark of cancers such as glioblastoma. The PKM2 gene variant is an important factor in the elevated aerobic glycolysis known as the Warburg effect in cancer, and miR-326 was demonstrated in glioblastoma to target PKM2.³¹ MiR-451 has been identified as a prognostic factor in glioblastoma and a regulator of glioblastoma metabolism and invasiveness.³² MiR-451 expression in glioblastoma is boosted by high glucose levels, and miR-451 directly targets the binding partner of liver kinase B1—an adenosine monophosphate kinase pathway protein activated in response to metabolic stress. Overexpression of miR-451 sensitized glioblastoma cells to glucose deprivation and inhibited migration, suggesting that downregulation of miR-451 helps glioma cells escape from metabolically stressful events or locations. This report also raised the prospect that individual microRNAs may promote or suppress malignant features of glioblastoma depending on context.

Immune Evasion

Exciting findings are emerging on glioblastoma escape from the immune response and on glioblastoma immunotherapy, and microRNAs are being recognized as having potential roles in both. Treatment of T cells isolated from glioblastomas with the tumor-suppressive miR-124 reversed a block in T-cell proliferation and also reduced expression of signal transducer and activator of transcription 3 and forkhead box P3—inhibiting development of immune-suppressive regulatory T cells.³³ MiR-124 delivery in mouse glioblastoma xenograft models prolonged survival, but only in immunocompetent mice. An earlier report demonstrated that miR-222 and miR-339 contribute to glioblastoma evasion of the immune system by targeting intercellular adhesion molecule 1, which modulates T-cell responses.³⁴

Glioblastoma Subtypes

While all glioblastomas share histopathological and clinical features, TCGA and other profiling efforts have revealed at least 3 glioblastoma subtypes.^35,36 The proneural subtype typically arises in frontal cortex, often has IDH1/IDH2 and TP53 mutations, and can demonstrate better prognosis and sensitivity to Notch inhibition.^35,37–39 The mesenchymal subtype is more aggressive, has greater vascularity, displays more frequent NF1 lesions, and may depend on TGF-β and NF-κB activity.^35,36,39 The classical subtype is aggressive and is marked by frequent EGFR lesions.^35,36,39 A possible fourth subtype, neural glioblastoma, is less well characterized. Individual microRNAs have been shown to be especially important in particular glioblastoma subtypes; examples include miR-34a for the proneural subset and miR-218 for the mesenchymal subset.^19,28 MicroRNA expression patterns in glioblastoma have themselves been used to establish an alternate grouping of glioblastoma subtypes. A recent analysis of microRNA expression profiles carried out using the TCGA glioblastoma dataset identified 5 subtypes that mimicked the microRNA profiles of different types of neural precursor cells: astrocytic, mesenchymal, multipotent, oligoneuronal, and neuronal.⁴⁰

MicroRNAs as Biomarkers

MicroRNAs have great promise as potential biomarkers in glioblastoma and other cancers. Numerous references have now indicated that levels of microRNAs in resected glioblastoma samples have prognostic import for patient survival. This has been shown for individual microRNAs such as miR-10b, miR-148a, miR-326, miR-181d, and others, but it has also been shown for multiple microRNA signatures.^41–44 Also exciting is the potential for quantitating microRNAs in patient serum and cerebrospinal fluid (CSF) samples, which can be sampled without surgery. Intriguing reports have shown that glioblastoma cells shed microvesicles with cytoplasmic contents including substantial quantities of microRNAs, and these microvesicles stably preserve microRNAs to allow quantitation in serum/CSF. This could allow relatively noninvasive microRNA biomarker detection for glioblastoma, and reports are emerging that investigate this.^45,46 This might also permit noninvasive determination of glioblastoma features such as subtype based on microRNA signatures. Interestingly, microvesicle shedding by glioblastoma cells enables them to “share” microRNAs with surrounding cells, modifying nearby stromal cells and essentially terraforming their environment.⁴⁷ This microvesicle sharing of microRNAs among glioblastoma cells may also have a critical therapeutic implication; while it is likely unrealistic to hope for 100% efficient delivery of microRNAs/inhibitors to a patient's glioblastoma cells, the glioblastoma cells that received a microRNA/inhibitor payload may well share it with nearby cells that did not.

MicroRNAs in Glioblastoma Therapy

Delivery of microRNAs or their inhibitors for effective glioblastoma therapy is an exciting possibility. However, despite numerous studies demonstrating oncogenic or tumor-suppressive functions for given microRNAs in glioblastoma, the published studies showing therapeutic efficacy in vivo are very limited. This is not surprising, given the therapeutic hurdle presented by the need for efficient delivery. Most reports have demonstrated an effect with ex vivo transfection of glioblastoma cells with a microRNA or its inhibitor before injection of the glioblastoma cells into mice subcutaneously or intracranially. However, a few reports have demonstrated in vivo efficacy of delivering microRNAs/inhibitors to established subcutaneous or intracranial glioblastoma xenografts. One recent report showed this for the tumor-suppressive miR-297 for established intracranial glioblastoma stem cell xenografts, and another study has shown this with systemic delivery of miR-124-3p.^33,48 In vivo delivery of microRNAs may be facilitated by lentivirus or by advances in nanoparticle delivery technologies. It is also likely to benefit from progress in local convection-enhanced delivery or in combining intravenous delivery with new methods to open the blood–brain barrier. Delivery of microRNAs or their inhibitors may also be effective in combination with other therapies, as a number of reports have shown that expression of certain microRNAs modulates glioblastoma sensitivity to radiation and chemotherapy.

MicroRNAs may also figure indirectly into novel therapeutic strategies for glioblastoma. In one study, it was found that incorporating target sites for downregulated microRNAs could be used to derive a glioblastoma-specific gene therapy.⁴⁹ In rare cases microRNAs may also lead to new targets in glioblastoma, as in recent reports describing that studies of miR-297 led to the identification of diacylglycerol kinase alpha as a promising target in glioblastoma.^48,50

Two reports in this issue make notable contributions to the field in light of the framework described above. In one, miR-124-5p—the opposite strand of the heavily studied miR-124-3p—targets laminin B1 to control glioblastoma angiogenesis. Notably, direct injection of this microRNA into established subcutaneous glioblastoma xenografts demonstrated therapeutic efficacy. The second study shows elegantly that miR-196a acts as an oncomiR in glioblastoma through targeting inhibitor of NF-κB alpha to upregulate the important NF-κB pathway. It provides a rare demonstration of therapeutic efficacy via delivering anti-microRNA therapy to established glioblastoma xenografts.

Funding

This work received no relevant funding.

Conflict of interest statement. None declared.