Circulating MicroRNA Biomarkers for Glioma and PredictingResponse to Therapy

Charlotte A. Tumilson & Robert W. Lea & Jane E. Alder &

Lisa Shaw

Received: 11 November 2013 /Accepted: 11 March 2014# Springer Science+Business Media New York 2014

Abstract The need for glioma biomarkers with improvedsensitivity and specificity has sparked research into shortnon-coding RNA known as microRNA (miRNA). AlteredmiRNA biogenesis and expression in glioma plays a vital rolein important signaling pathways associated with a range oftumor characteristics including gliomagenesis, invasion, andmalignancy. This review will discuss current research into therole of miRNA in glioma and altered miRNA expression inbiofluids as candidate biomarkers with a particular focus onglioblastoma, the most malignant form of glioma. The isola-tion and characterization of miRNA using cellular and molec-ular biology techniques from the circulation of glioma patientscould potentially be used for improved diagnosis, prognosis,and treatment decisions.We aim to highlight the links betweenresearch into miRNA function, their use as biomarkers, andhow these biomarkers can be used to predict response totherapy. Furthermore, increased understanding of miRNA inglioma biology through biomarker research has led to thedevelopment of miRNA therapeutics which could restorenormal miRNA expression and function and improve theprognosis of glioma patients. A panel of important miRNAbiomarkers for glioma in various biofluids discovered to datehas been summarized here. There is still a need, however, tostandardize techniques for biomarker characterization to bringus closer to clinically relevant miRNA-based diagnostic andtherapeutic signatures. A clinically validated biomarker panelhas potential to improve time to diagnosis, predicting responseto treatment and ultimately the prognosis of glioma patients.

Keywords MicroRNA .Glioma . Biomarker . Serum .

Cerebrospinal fluid

Introduction

Biomarkers are defined as objectively measured characteris-tics within the body which are used to gain information abouta particular disease [1]. Early detection and effective treatmentstrategies for glioma patients are vital for improving clinicaloutcomes, and developing biomarkers for this purpose haslong been an aim of research. In addition, biomarkers canprovide an insight into the characteristics of the neoplasm.Biomarkers are either produced by the pathological processesof tumor progression or by the host system in response to thetumor [2]. The information that biomarkers provide aboutcancer can be used to predict important factors such as prog-nosis [3, 4] and response to therapy [5], as well as to improvediagnosis [6] and to assist earlier detection [2]. Biomarkerscan also be used to differentiate between different tumorgrades [1] and subtypes [7], both of which can be used totailor treatment strategies.

Current cancer biomarkers in clinical use are outlined inTable 1. Although these biomarkers are used for diagnosis andtreatment, they lack sufficient sensitivity and specificity re-quired for a successful biomarker [8]. Current diagnosticbiomarkers, such as prostate specific antigen (PSA), are sub-ject to a high incidence of false positive diagnoses [9]. Mucin-16, a serum-based diagnostic marker of ovarian cancer, lackssensitivity and specificity for early diagnosis because 50 % ofpatients in the early stages of ovarian cancer do not presentwith serum mucin-16 expression [10]. Alpha-fetoprotein andbeta-human chorionic gonadotropin, both diagnostic markersfor testicular cancer, are found to be up-regulated in only 60%of patients therefore risking a false negative diagnosis [11].While proving to be an effective biomarker for breast cancer,the up-regulation of HER2 occurs in only 20 to 30% of breasttumors making it an effective biomarker for only a smallpopulation of patients [12]. There is need for biomarkerswhich can be identified in the majority of a patient populationand with a greater sensitivity to reduce the risk of incorrect

C. A. Tumilson :R. W. Lea : J. E. Alder : L. Shaw (*)School of Pharmacy and Biomedical Sciences, University of CentralLancashire, Preston, Lancashire PR1 2HE, UKe-mail: LShaw1@uclan.ac.uk

Mol NeurobiolDOI 10.1007/s12035-014-8679-8

diagnosis. One such family of molecules which promisesincreased sensitivity and specificity as a biomarker ismicroRNA (miRNA). MiRNA are small non-coding RNAmolecules that are normally around 22 nucleotides in length.MiRNA function to regulate cell behavior [13] and modulategene expression at the post-transcriptional level, by binding tomRNA and suppressing translation [14].

The focus of this review will be circulating miRNAbiomarkers for glioma and how they can be furtherresearched to aid the successful treatment of this disease.This includes the current limitations faced by researchersattempting to discover new, effective biomarkers and treat-ments. Glioma is the most common primary brain tumorthat arises from glial cells within the central nervous system(CNS). In adults, 60 % of these are malignant forms suchas anaplastic astrocytoma and glioblastoma multiforme[15]. In malignant cases, 70 % are glioblastomamultiforme, the most aggressive and invasive type of glio-ma [16]. The prognosis for patients with high-grade gliomais poor; patients with anaplastic astrocytomas have a sur-vival rate of 2–3 years with surgical radiotherapy andsometimes chemotherapy treatment [17]. Patients with glio-blastoma multiforme have the least promising prognosis;even after aggressive treatment including maximal surgicalresection and chemotherapy, recurrence is common andsurvival is only 12–15 months [18].

The difficulty in identifying a specific biomarker for glio-ma lies partly in the complex heterogeneous nature of thecancer itself. The multiple mutations a tumor cell undergoesduring transformation and the frequency of genomic changesbetween grades and sub-types within the grades all contributeto this heterogeneity. Using a group of biomarkers to detect arange of these characteristics or a set of related biomarkers forone specific characteristic is more beneficial than using asingle biomarker. This has driven research to the identificationof multiple biomarkers which could be used together in apanel [19]. These panels could be detected using a range ofreadily accessible high-throughput techniques including quan-titative real-time polymerase chain reaction (qRT-PCR) [20,21] and ELISA [22]. The main appeal of miRNA biomarkersis the tissue and cell specificity of their expression which canbe illustrated by miR-10b [23]. MiR-10b has been identified

in both low- and high-grade glioma but is not present in non-cancerous brain tissue. As well as demonstrating tissue spec-ificity in their expression profiles, miRNAs are known tocontribute to the progression of glioma. Through loss-of-function studies, miR-10b was implicated in the regulationof glioma proliferation and apoptosis. In these studies, theoverexpression of miR-10b caused the up-regulation of cellcycle regulators including cyclin B1 and D1. Up-regulation ofmiRNAs usually causes a down-regulation of direct miRNAtargets rather than an up-regulation as observed in this study,which led to the suggestion that miR-10b promotes gliomacell growth by indirectly influencing cell cycle regulators [23](Fig. 1). The inhibition of miR-10b led to senescence, growtharrest, and apoptosis, both in vitro and in vivo [23]. The role ofmiRNA in glioma development and progression, and theirspecificity, makes them important candidate biomarkers thatcould provide important characteristic information of a tumorand improve treatment and prognosis.

MicroRNA Profiles in Glioma

In 2005, the first miRNA profiles were obtained from a rangeof cancers using bead-based flow cytometry. MiRNA expres-sion profiles were found to differ across cancer types, but ageneral decrease in miRNA expression was observed in allsamples. Profiling was also able to discriminate between thedevelopmental origin of samples [24]. This could be particu-larly useful in metastatic cancers to determine the tissue oforigin and location of the primary neoplasm [25]. Differentmutations within the same cancer type, demonstrated by dif-ferences in miRNA expression, were also observed within thisstudy [24]. These differences can affect response to treatmentand prognosis [26], and to identify these mutations usingspecific miRNA profiles would permit better understandingof the tumor and enable improved treatment decisions andpatient outcomes [26].

MiRNA signatures have been identified in both glioblas-toma tissue and the circulation of glioblastoma patients. Re-cently, the employment of deep sequencing produced one ofthe largest sets of miRNA profiles for glioblastoma and con-trol brain tissue [27]. This study identified 33 up-regulated

Table 1 Current biomarkers andtheir applications Biomarker Cancer type Type of biomarker

Prostate-specific antigen (KLK3) Prostate cancer Diagnostic

Mucin-16 Ovarian cancer Diagnostic

Alpha-fetoprotein Testicular cancer Diagnostic

Beta-human chorionicgonadotropin

Testicular cancer Diagnostic

Her-2 Breast cancer Prognostic and risk of recurrence

PTK7 T cell acute lymphoblastic leukemia Response to treatment

Mol Neurobiol

miRNA in the glioblastoma tissue and 40 down-regulated. Inaddition, 18 novel miRNAs and 16 novel miRNA-3ps wereidentified (Table 2). Moreover, this study found that for 23miRNA, the most abundant isomiRs, miRNA isoforms, werethose with 5′ variations. The repository of large datasets suchas those in the aforementioned study provides a valuableresource for biomarker identification without the need forsample collection.

Increasing the glioblastoma sample size by utilizing multi-ple studies from public databases increases the statisticallikelihood that the miRNA signature found will be clinicallyrelevant. One such study from The Cancer Genome Atlas(TCGA) public repository analyzed individuals with glioblas-toma and identified an expression signature of ten miRNAswhich could be used to predict survival. Three up-regulatedmiRNAs—20a, 106a, and 17-5p—in the signature were asso-ciated with a better survival rate. The remaining seven—31,222, 148a, 221, 146b, 200b, and 193a—were associated withpoor survival. It was concluded that these miRNAs could also

be used as therapeutic targets for the treatment of glioblastomain addition to being used as prognostic markers. [28]. Our ownunpublished data shows that four of the miRNAs identified inthis study are also up-regulated in the serum of glioma patientsin our study. These are miR-20a, 17-5p, 222, and 148a. Ourdata also suggests differences in expression between agegroups, showing variability in miRNA expression betweendifferent patient populations. Analysis of the TCGA datashowed 19 miRNAs in glioblastoma with gender, race, ther-apy, and recurrence-dependent differences. These miRNAsincluded miR-222, 181, 34, and 140. These findings showthat there are a range of population variables which couldaffect miRNA expression and therefore biomarker panel se-lection. Taking these variables into account will increase thesuccess of such panels as well as their specificity toward theirintended populations [29].

MiRNA profiles have shown that glioblastomas are madeup of subclasses which vary in clinical and genetic character-istics. Analysis of profiles from the TCGA identified fivesubclasses relating to neural precursor types. Identificationof sample subclasses in a study could provide subtype-specific miRNA biomarkers. Furthermore, additional infor-mation of patient characteristics within the subtypes such asage and response to treatment could be extrapolated to predictsurvival. MiRNA markers identified for these characteristicsincluded miR-222 for gender, miR-137 as a race-dependentmarker, and miR-140 as a marker for survival [30]. Anotherstudy analyzing the TCGA dataset has demonstrated thisapproach and identified subtype-specific prognostic miRNAbiomarkers which were subsequently validated using qRT-PCR. Each subtype displayed a panel of biomarkers which

Table 2 A total of 16novel miRNA-3p identi-fied in glioblastoma andnon-cancerous tissue

Novel miRNA-3p discovered

miR-539 miR-758

miR-382 miR-98

miR-1307 miR-873

miR-212 miR-135a-2

miR-204 miR-511-1

miR-301a miR-1271

miR-181b-1 miR-381

miR-3676 miR-487a

Fig. 1 Relationship between miRNA and glioma characteristics. Altered expression of miRNA affects a number of different targets which influence thesame glioma characteristic

Mol Neurobiol

contained both positive and negative prognostic markers,some of which were shared by more than one subtype [31].The classical subtype of GBM displayed seven prognosticmiRNA, five risky—miR-26a, 767-3p, 153, 31, and 222—and two protective—miR-654 and 422b. Identification of apatient’s glioblastoma subtype will provide important infor-mation on response to treatment and prognosis as well asproviding a diagnosis.

Role of microRNAs in Glioma

Research into the effect of dysregulated miRNA and their rolein glioma progression has identified their importance withinpivotal signaling pathways. These signaling pathways areimportant when determining glioma characteristics which af-fect prognosis and treatment outcomes and provide potentialtargets for biomarker development. Understanding the role ofmiRNA in signaling pathways and how they affect progressionpermits the identification of miRNA biomarkers for specificpurposes, for example diagnosis or treatment choice. Thissection outlines the role of miRNA in such pathways whichaffect glioma behavior and contribute to a range of character-istics including migration, invasion, and proliferation.

Since the discovery of the first miRNA in 1993 [32], thebiogenesis and maturation of miRNA has been well-researched [33]. The basic process of miRNA biogenesis in-volves transcription of the miRNA gene to produce a primarymiRNA (pri-miRNA) which is processed into a stem-loopprecursor miRNA (pre-miRNA). Cleavage of the pre-miRNAproduces a miRNA duplex which is again cleaved into 5p and3p mature miRNA strands, named in relation to the orientationof the seed sequence. The mature miRNA strand which is notsubsequently incorporated into the RISC complex is labeled asmiRNA* (Fig. 2). Research is now beginning to identify newlevels of complexity in miRNA regulation, and current studiesare showing that deregulation of biogenesis and maturation canalso contribute to tumorigenesis.

The biogenesis of miRNA is known to be subject to regu-lation by RNA editing. Pri-miRNA transcripts can undergoconversion of adenosine to inosine by adenosine deaminaseacting on RNA (ADAR), known as A to I editing [34]. A to Iediting has been found to affect pri-miRNA processing andhas been shown in some studies to result in a reduction inlevels of mature miRNA [35]. Alternatively, editing can alsolead to the production of mature miRNA with altered se-quences, known as “isomiRs,” which target different mRNAtranscripts [34] and can contribute to tumor progression. Ev-idence suggests that miRNA transcripts in glioma, particularlyhigh grade, undergo A to I editing as a result of dysfunctionalADARs [36]. Editing within the seed sequence can alter thetarget of a miRNA. This disrupts regulation of protein expres-sion and contributes to progression [36]. On the other hand,

lack of editing can also lead to altered gene and proteinexpression. One such study identified reduced editing of themiR-376 cluster in glioblastoma as a result of lower expres-sion of ADAR and the isoform ADARB1. This caused accu-mulation of unedited miR-376a-3p transcripts and was shownto contribute to invasiveness and migration of glioma cells[37] (Fig. 1). The low expression of ADAR could potentiallyaffect other, currently unknown, targets which may also con-tribute to the progression of high-grade glioma [37]. Anothersource of alternative miRNA transcripts is the cleavage ofmiRNA duplexes during maturation. The duplexes are cleavedto produce a functional mature miRNA which is incorporatedinto the RISC and a miRNA* believed to be degraded follow-ing cleavage (Fig. 2). One such study, however, found thatmiRNA* transcripts are functional and capable of translationalrepression of mRNA targets, which could therefore have anotable effect in the pathology of disease [38].

Although miRNAs have different targets in different sig-naling pathways, their effects can all lead to one commontumor characteristic (Fig. 1). For example, miR-23b, miR-130b, and miR-107 regulate different signaling pathways butall contribute to invasion. This underlines the complexity andimportance of miRNAs in glioma biology, and altered

Fig. 2 Cleavage and incorporation of miRNA and miRNA* into RISCcomplex. Pre-miRNA is cleaved by DICER to produce a predominantmature miRNA strand and a complementary miRNA* strand, both ofwhich can be incorporated into RISC and function as post-transcriptionalregulators

Mol Neurobiol

expression signatures could again provide important informa-tion characteristic of a particular tumor. A recent study into therole of miR-107 in glioma migration and invasion identifiedNotch2 as a key target [39]. MiR-107 is down-regulated inglioma, and over-expression in glioma cell lines led to down-regulation of Notch2, which controls a number of tumorcharacteristics including migration (Fig. 1). The authors con-cluded that down-regulation of miR-107 in glioma promotesmigration and invasion through Notch2 signaling pathways[39]. A comparison of miRNA expression profiles in migra-tory and migration restricted groups of glioblastoma cell linesidentified miR-23b as a regulator of both migration and inva-sion [40]. MiR-23b down-regulation results in increased ex-pression of the non-receptor tyrosine kinase Pyk2 which leadsto cell migration and invasion (Fig. 1). It was concluded thatidentifying miRNA important in regulating glioma invasioncould provide targets for modulation to reduce invasion andimprove treatment outcomes of glioma [40]. Unpublished datafrom our own laboratory have identified miR-23a as a miRNAwith altered expression in the serum of glioblastoma patients.MiR-23a shares sequence similarities to 23b and has beenfound to be regulated by cAMP response element-binding

protein (CREB) and through increased expression promotesglioma cell growth and survival [41] (Fig. 1). An isoform ofp63 containing a transactivation domain known as TAp63 isknown to be structurally and functionally similar to the tumorsuppressor protein p53. TAp63 has been identified as a tumorsuppressor which repressesmetastasis. In a study involving bothmouse and human tumor cell lines, Dicer and miR-130b wereshown to be targets of TAp63. Binding of TAp63 to the Dicerpromoter led to transcriptional activation. As well as regulationof Dicer, TAp63was found to target miR-130b leading to its up-regulation and decrease in invasion (Fig. 1). Inactivation ofTAp63 leads to increased metastasis and invasion [16].

Dysregulation of the transforming growth factor β(TGF-β)/smad pathway in high-grade glioma is known tocontribute to tumor progression [42]. MiR-182 has beenshown as a target of TGF-β which, once activated, down-regulates the expression of NF-κB inhibitors (Fig. 3). Onesuch inhibitor is ubiquitin carboxyl-terminal hydrolase (CYLD)which is known to be expressed at a lower level in glioma.Although CYLD is down-regulated in glioma, CYLD mRNAlevels do not alter, suggesting regulation by direct targeting bymiR-182 [42] (Fig. 1). NF-κB and the associated signaling

Fig. 3 Role of miR-182 in the NF-κB pathway and miR-30e* in thenegative feedback loop adapted from KEGG. Constitutive activation ofthe TGF-β/Smad pathway increases miR-182 expression. MiR-182 re-duces inhibition of NF-κB intermediary signaling molecules by binding

to CYLD mRNA causing increased activation of the pathway. MiR-30e*binds to IKβα mRNA preventing action of the negative feedback loopand again, increasing the activity of the NF-κB pathway, and therebycontributing to aggressiveness in glioma

Mol Neurobiol

pathways are constitutively activated in cancer including glio-ma and contribute to tumorigenesis [43]. The inhibition of theNF-κB negative feedback loop by over-expression of miR-30e* constitutively activates this signaling pathway in glioma[44] (Figs. 1 and 3). Although further research is needed toelucidate the full function of miR-182 in this pathway, miR-182may also inhibit the negative feedback loop and either miRNAmay function individually or together to sustain NF-κB signal-ing in cancer cells and tumorigenicity in glioma.

To sustain hyperproliferation within a tumor, cells mustadapt their metabolic pathways to utilize energy sources avail-able to them both in hypoxic and normoxic conditions. Ourresearch group has identified miR-451 as a miRNA withaltered expression in the serum of glioma patients (unpub-lished data). MiR-451 has been found to regulate pathwaysinvolved in the adaptation of cancer cells to metabolic stress.MiR-451 is a regulator of the liver kinase B1/Amp-activatedprotein kinase (LKB1/AMPK) signaling pathway. Down-regulation of the pathway in the absence of glucose permitsthe activation of LKB1, allowing cancer cells to adapt tocellular stress by reducing proliferation and activating migra-tory processes [45]. Furthermore, down-regulation of miR-451 in glioblastoma patients is associated with reduced sur-vival, providing a potential prognostic biomarker [45] (Fig. 1).

Circulating MicroRNA

To prevent degradation in the circulation, miRNAs are re-leased by cells in both exosomes and miRNA/protein com-plexes. Exosomes are lipid vesicles ranging between 50 and100 nm in size and contain a range of molecules includingmRNA, miRNA, DNA, and proteins. The detection of bio-markers within serum is attractive due to the relatively non-invasive process of collection [46], and subsequent isolationof miRNA from the circulation is a convenient method forbiomarker detection.

Exosomes originate from multi-vesicular bodies (MVB)within the cell which fuse with the cell membrane through asecretion pathway involving Rab GTPases to release theexosomes into the extracellular environment. Analysis of thelipid membranes of exosomes has shown them to be enrichedin the sphingolipid ceramide suggesting it to be a pivotalcomponent of exosome formation. It has been shown thatlipid-raft microdomains containing high levels ofsphingomyelin promote the inward budding of MVBs follow-ing the production of ceramide from sphingomyelin, to formexosomes [47]. Silencing of two members of the Rab GTPasefamily, Rab27a and Rab27b, resulted in a reduction of MVBdocking to the cell membrane in the HeLa cell line [48],showing that Rab27a and Rab27b and their effector proteinscan promote intracellular trafficking ofMVBs and subsequentexosome secretion in certain instances. Further to this study,

the authors investigated the function of Rab27a and Rab27b inin vivo murine breast carcinomamodels. Although they foundthat Rab27a modulates exosome secretion as shown in theprevious study, Rab27b on the other hand was found not to berequired for exosome secretion in the murine models. Thesefindings suggest that MVB trafficking and secretion is acomplex pathway which may not be identical across cell typesand therefore must be further elucidated [49].

Exosomes are known to be released by both non-cancerousand cancerous cells [50] as a form of cellular communicationand perform a variety of functions depending on their contentsand cellular context. Exosomes originating from different celltypes share a standard set of proteins including tetraspanins andheat shock proteins [50], as well as proteins specific to their cellof origin such as the tumor-specific EGFRvIII [51]. Although itis unclear as to how miRNA, mRNA, and proteins are pack-aged into exosomes, research has shown that the packaging ofexosomes is a specific and selective process [50]. Only certainmiRNAs are incorporated and released into the circulation dueto the selectivity of exosome packaging. As a result, certainderegulated miRNAs within glioma cells may not be presentwithin isolated exosomes, and therefore, tissue biomarkers maynot translate into circulatory biomarkers. This leads to theconclusion that miRNA signatures for tissue and circulatorybiomarkers need to be investigated independently.

Analysis of primary glioblastoma tissue and exosomes byqRT-PCR identified the presence of 11 miRNAs in bothsample types. These 11 miRNAs such as miR-21 were report-ed to be commonly up-regulated in glioma. These miRNAswere shown to be present in the exosomes at a lower level thanin the corresponding glioblastoma tissue but showed goodcorrelation with the tissue profile leading to the conclusionthat circulating exosomes could provide a “snapshot” on theglioblastoma transcriptome [51]. MiR-21 was also identifiedin exosomes derived from the U251 glioblastoma cell line. Inthis study, over 28 miRNAs were identified in U251exosomes, and at least 22 were significantly enriched inexosomes isolated from culture media compared to cell lineexpression suggesting specific tumor modulatory roles [52].Further to this, not only was there an enrichment of certainmiRNAs, there was also a higher level of nine miRNA*including miR-181a*, 93*, 452*, and 106a* compared tothe mature miRNA form as well as an increase in 3p miRNAwhen the 5p form was also present. The authors concludedthat this change in abundance of the less dominant miRNAscould result in different mRNAs being targeted within recip-ient cells than those in the glioma cells.

Compared to other biomarker types, detection of a smallpanel of miRNA from the circulation using techniques such asqRT-PCR provides increased sensitivity. The required startingconcentration of total RNA is relatively low, as little as 25 pgof RNA is required from biofluid samples for biomarkerdetection [53]. The selective packaging of miRNA into

Mol Neurobiol

exosomes containing components indicating their cell of ori-gin and the detection of tissue-specific miRNAs in the circu-lation provides specificity for miRNA biomarkers. In addi-tion, the detection of more than one tissue-specific miRNAprovides the advantage of reducing overlap with other pathol-ogies which may share deregulated miRNA biomarkers [54].

Solexa sequencing of pooled sera identified a panel ofseven down-regulated circulating miRNAs which could beused as a signature for glioma [46]. The seven miRNAs whichmade up this panel included miR-15b*, miR-23a, miR-133a,miR-150*, miR-197, miR-497, and miR-548b-5p. In additionto this main panel, certain miRNAs made up smaller groupswhich could be used to differentiate between benign andmalignant astrocytomas, as well as other primary brain tumors[46]. The expression of miR-15b*, miR-23a, miR-150*, miR-197, and miR-548b-5p was significantly up-regulated in ma-lignant neoplasms in comparison to benign samples.

A less studied source of miRNA biomarkers is plasma; onesuch study showed that the levels of certain miRNAs includ-ing 21, 128, and 342-3p were altered in the plasma of glio-blastoma patients in comparison to non-cancerous plasmasamples [55]. This particular study used qRT-PCR to identifytarget miRNAs from individual plasma samples which couldaccount for differences between the miRNAs identified bySOLEXA sequencing, as well as other variables including asmaller sample size, different extraction, and qRT-PCR re-agents being used. This study not only identified potentialmiRNA biomarkers but also showed that isolation of miRNAneed not be restricted to serum alone.

Due to the close proximity of cerebrospinal fluid (CSF) tothe brain and spinal cord, disorders arising in the CNS canoften cause an alteration in CSF composition. The presence ofa glioma within the CNS results in the alteration of CSFcomposition as a result of: humoral responses [19], break-down of structures within the CSF such as the blood–brainbarrier (BBB) [56], or as a result of up-regulated productionand secretion by the glioma cells themselves [57]. Alterna-tively, the function of the structures related to CSF productionand composition can become affected and subsequently con-tribute to pathophysiology [19]. The role of CSF in the path-ogenesis of glioma is mainly the delivery of substances thatplay a role in tumorigenesis. These substances include growthfactors, hormones and signaling molecules, as well as manyother components of CSF and are believed to contribute to anumber of glioma characteristics such as invasion, migration,and metastases [19].

The presence of miRNA in the CSF of glioma patients hasinitiated studies into its potential as a source of biomarkers.Although CSF is not routinely obtained from patients withglioma [58], the proximity of CSF to a glioma, and its isola-tion from general circulation, means it could provide a morespecific and accurate miRNA profile in comparison to serumand plasma [59]. In 2012, the identification of miR-21 and

15b within the CSF of patients with malignant glioma wasreported. MiR-21 promotes migration and invasion bytargeting MMP inhibitors RECK and TIMP3 [60]; miR-15bis a cell cycle regulator which targets CCNE1 [61] (Fig. 1).This initial study highlighted CSF as a potential source formiRNA biomarkers and concluded that in the future, CSFmiRNA could differentiate between glioma subtypes [14].Following this, a pilot study [59] identified miRNA whichcould potentially be used to diagnose glioblastoma or discrim-inate between glioblastoma and metastatic cancer, once againhighlighting the potential of CSF as a source of biomarkers forglioma. Both studies highlighted miR-21 in the CSF of glio-blastoma patients, but the latter study also found miR-10bsignificantly up-regulated. Furthermore, members of the miR-200 family, which share the same seed sequence to each other,were up-regulated in the CSF of patients with brain metastasis.A member of the miR-200 family, miR-200b, targets CREB1and regulates glioma growth thereby acting as a tumor sup-pressor (Fig. 1) [62].

From our own unpublished data, we have identified anumber of miRNA with altered expression in the serum andCSF of glioma patients. Some of these include miRNAshighlighted in this review which have been linked togliomagenesis and function including miR-17-5p, 20a, 21,148a, 222, and 451. Collating the data from this review, wepropose a glioma-specific miRNA panel for detection ofcirculating miRNA in biofluids (Table 3) and summarize themiRNA mentioned here and their targets (Table 4).

Table 3 Candidate miRNA biomarker panel constructed from currentresearch of various biofluids

Serum CSF Plasma

15b* down-regulated 15b up-regulated

17-5p up-regulated

20a up-regulated

21 up-regulated 21 up-regulated

23a down-regulated

31 down-regulated

106a up-regulated

128 down-regulated

146b up-regulated

148a up-regulated

150* down-regulated

193a up-regulated

197 down-regulated

200b up-regulated 200 up-regulated

221 down-regulated

222 down-regulated

342-3p down-regulated

548b-5p down-regulated

Mol Neurobiol

MicroRNA Biomarkers for Response to Therapy

In the treatment of glioma, chemoresistance can be a pivotalfactor in the prognosis of a patient. The ability to predictresponse to treatment could improve prognosis by selectingthe right treatment course as soon after diagnosis as possibleand permit rapid adaption of treatment to the acquisition ofchemotherapeutic and radio-resistance, a common problem inthe treatment of glioblastoma patients. This not only benefitsthe patient by improving their prognosis but also improves thecost–effectiveness of chemotherapeutics by using them onlywhen they are expected to succeed. Furthermore, the use ofpredictive biomarkers in clinical trials could identify patientsmost likely to respond to new anti-cancer therapies, therebyaccelerating the development of novel therapeutics [63]. Cur-rently, the gold standard for glioblastoma treatment is temo-zolomide (TMZ) usually combined with radiotherapy. Only asmall subset of patients respond to TMZ treatment, as patientswith the functional O6-methyl guanine methyltransferase(MGMT) DNA repair protein reverse the guanine methylationcaused by TMZ leading to chemoresistance and limited

success of this drug. MiR-181d could be used as a biomarkerto identify patients who would respond the best to TMZ.MGMT is a candidate target of miR-181d, and a higherexpression of miR-181d correlates with a lower expressionof MGMTand subsequently improved response to TMZ [64].

While miR-181d up-regulation may correspond to a betterresponse to TMZ, up-regulation of miR-21 on the other handmay predict poor response. This poor response relates to theproblem of the high rate of TMZ resistance which develops inpatients. MiR-21 is one of the most frequently up-regulatedmiRNAs in glioblastoma and has been found to protectU87MG glioblastoma cells from TMZ-induced apoptosis[65]. Inhibition of miR-21 in the D54MG cell line enhancedchemosensitivity to TMZ following the induction of resis-tance [66]. These findings both suggest that miR-21 couldbe used as a biomarker to predict or monitor the acquisitionTMZ resistance in glioblastoma patients to enable quick ad-aptation in treatment strategy and maintain a good prognosis.Further to its role as a chemotherapeutic marker, miR-21 hasalso been shown to function in the acquisition of radio-resistance. Analysis of radio-resistance in a number of

Table 4 Deregulated miRNAs from different sample types in glioma and their corresponding signaling pathways

MicroRNA Up-/down-regulation Targets/signaling pathway Sample type

10b Up BCL2L11/Bim, CDKN1A/p21, CDKN2A/p16 Tissue, cell lines, CSF

15b* Down – Serum

15b Up CCNE1 CSF

17-5p Up Cyclin D1 Tissue

20a Up E2F1, cyclin D1 Tissue

21 Up PI3k/Akt Tissue, serum, plasma, CSF

23b Down Pyk2 Cell lines

23a Down IL6R Serum

30e* Up NF-κB Tissue, cell lines, primary culture

31 Down FIH Tissue

106a Up FASTK Tissue

107 Down Notch2 Tissue, cell lines

128 Down E2F3a Plasma

133a Down CAV1, LIM, LASP1 Serum

146b Down MMP Tissue

148a Down DNMT1 Tissue

150* Down – Serum

182 Up TGFβ/smad, NF-κB Tissue, cell lines, primary culture

193a Down Mcl-1 Tissue

197 Down Fus1 Serum

200b Down RND3 Tissue

221 Down PTEN, p27, and p57 Tissue

222 Down PTEN, p27, and p57 Tissue

342-3p Down BMP7 Plasma

451 Down LKB1/AMPK Cell lines

497 Down BCL2 Serum

548b-5p Down PTEN, CDK6 Serum

Mol Neurobiol

glioblastoma cell lines including U87MG and U373 showedthat radio-sensitivity was closely related to the expressionlevel of miR-21 [67]. The silencing of miR-21, using anti-miR-21 in radio-resistant malignant glioma cell lines, led tothe sensitization of these cells to radiation [67]. Anti-miR-21was found to sensitize U87MG and U373 cells through inac-tivation of the PI3K/Akt signaling pathway. While these find-ings point toward a mechanism of acquired radio-resistance,they also highlight that miR-21 levels could be an importantpredictor of acquired radio-resistance. The use of which, onceagain, would permit quick adaptation of treatment plans andeffective treatment of the glioblastoma as it progresses.

As well as affecting TMZ and radio-resistance, miR-21along with miR-30b and 30c have been identified as regula-tors of TNF-related apoptosis-inducing ligand (TRAIL)-in-duced apoptosis. These three miRNAs therefore could affectthe sensitivity of glioma cells to treatment with the TRAILligand [68]. A study of glioma cell lines and primary culturesfound that miRNA-21, 30b, and 30c were significantly up-regulated in TRAIL-resistant glioma cell lines [68]. Conse-quently, TRAIL-sensitive cells were found to exhibit down-regulation of these three miRNAs. The targets of thesemiRNAs were found to include caspase-3 by the miR-30family and Tap63 by miR-21. It was concluded that thesemiRNAs regulate apoptotic programs within glioma cell lines.The results obtained also demonstrated a difference in miRNAexpression between TRAIL-sensitive and TRAIL-resistantcells [68]. These miRNA biomarkers for response to treatmentand therapeutic resistance are outlined in Table 5.

Vascular endothelial growth factor (VEGF) up-regulationcan be used as a biomarker to predict early recurrence followingchemoradiation treatment. Analysis of VEGF expression in theurine of glioblastoma patients has been shown to be a markerfor 1-year progression-free survival [69]. The identification ofVEGF in the urine of glioma patients highlights the possibilityof other deregulated tumor-associated factors, such as miRNA,being isolated and used as biomarkers in this biofluid. To ourknowledge, there are no current studies published relating to theidentification of miRNA biomarkers in the urine of gliomapatients; however, the measurement of miRNA biomarkers inurine would provide a completely non-invasive test.

As well as contributing to gliomagenesis and treatmentresistance, miRNAs can also be used to combat gliomagrowth as a novel treatment. The manipulation of deregulatedmiRNA levels in glioma, using methods where the comple-mentary sequence can act as a sponge or a synthetic mimic[70], could potentially be used as effective treatments [71];therefore, research into biomarker identification and the de-velopment of related methods can assist the development ofnovel therapies for glioma. In cancers, including glioma,expression of tumor suppressor miRNA is down-regulated,and therefore, it is an aim of research to restore the function ofthese miRNA. A miR-34 mimic known as MRX34 is current-ly undergoing testing in a phase I clinical trial for the treatmentof primary liver cancer. MiR-34 is a tumor suppressor whichregulates 24 oncogenes which affect proliferation, metastasis,and chemoresistance among other tumorigenic properties.MRX34 is designed to restore the tumor suppressor effect ofmiR-34, and phase I trials will determine off-target and dosageeffects of the mimic [72]. Like primary liver cancer, miR-34 isdown-regulated in glioblastoma; therefore, the success of thismimic as a treatment for liver cancer may lead to its use as anovel therapy in other neoplasms. Another novel treatment totarget down-regulated miRNA in glioma is the small moleculeEnoxacin. Enoxacin binds TAR RNA-binding protein 2(TRBP2), a protein involved in miRNA biosynthesis, andincreases the production of miRNAs including tumor suppres-sors. The use of this therapy restores down-regulated tumorsuppressing miRNAs in cancer cell lines and in vivo mousemodels and causes cancer cell specific growth inhibition [73].

Challenges in Biomarker Discovery

Although miRNAs appear to be extremely promising bio-markers and research into the identification of miRNA bio-markers for cancer remains on the increase, to date, there areno clinically utilized miRNA biomarkers for glioma. Ongoingclinical studies investigating miRNA profiles for a number ofcancers have been outlined in a previous review [71]. Re-searchers believe that the lack of clinical miRNA biomarkerscompared to the number identified in research is due tolimitations in standardizing of sample type collection [74],determining optimal methods of extraction, and processing ofboth samples [75] and data [76], all of which can affect thereproducibility of individual findings.

The choice of sample type, and the origin of miRNA, canbe a limitation in biomarker discovery. An analysis of circu-lating miRNA isolated from plasma samples identifiedmiRNAs of hematopoietic cell origin. The presence of thesemiRNAs in plasma samples is a pre-analytical variable whichcould affect the analysis of circulating miRNA expression.The employment of sub-fractionation to remove cellularmiRNAs improves specificity of circulating miRNA markers.

Table 5 MicroRNA biomarkers for response to therapy

MiRNA Therapeutic response

MiR-181d Temozolomide resistance

MiR-21 Temezolomide resistance, TRAIL resistance,Radio-resistance

MiR-30b TRAIL resistance

MiR-30c TRAIL resistance

MiR-425-5p Radiochemotherapy response

MiR-93-5p Radiochemotherapy response

Mol Neurobiol

The use of miRNA categories for the classification of circu-lating and cellular miRNA in plasma samples was proposed asa method that could improve miRNA biomarker studies [77].The expression of specific miRNAs can differ between sampletypes used in studies [55].While miR-15b did not appear to besignificantly dysregulated in a study using plasma samples, asignificant increase in miR-15b levels in CSF samples hasbeen reported [14]. Our own laboratory has also identifiedmiR-15b-5p as a miRNAwith altered expression in serum andCSF samples (unpublished data). Initial collection of samplebiofluids must also be taken into account as there are variablesin this process that may also affect miRNA data. Differencesin collection tube type and phlebotomy techniques have bothbeen suggested as factors which may cause variability incirculating miRNA expression. Hemolysis of the sample canincrease the abundance of certain miRNA in biofluid sampleswhich can subsequently affect biomarker selection [75]; there-fore, effective removal of cells is essential when using plasmaand serum samples for biomarker detection. In addition tocells, other components of serum and plasma which are pres-ent in high levels, such as lipids and proteins, can affect theisolation of the RNA [78]. Although tumor cells are known toshed exosomes, non-neoplastic cells and platelets also releaseexosomes. Serum isolated exosomes were used to identifymRNA expression patterns in glioblastoma patients [79].When compared with control serum samples, the expressionpatterns of mRNA within the exosomes could differentiatebetween those patients with glioblastoma and those without;however, the down-regulation of mRNA in serum exosomescould either be due to tumor exosomes or from exosomes withan altered expression due to disease states and lifestyle fromnon-cancerous cells [79]. Like mRNA expression, the releaseof miRNAs from non-neoplastic cells must be taken intoaccount in circulating miRNA expression studies for the iden-tification of biomarkers [80].

The tissue and cell specificity of miRNA and individualvariations due to diet, disease, infection, or even age andgender could potentially affect the detection of circulatingmiRNA biomarkers. One such review illustrated the influenceof a number of environmental factors including diet, infection,and stress on epigenetic mechanisms including miRNA ex-pression [81]. Although the focus was the effect of thesefactors on cancer susceptibility, it also shows that inter-individual variations in lifestyle can affect the expression ofmiRNA which could subsequently affect their detection anduse as biomarkers [81]. The analysis of gene expression in theperipheral blood of healthy subjects also demonstrated inter-individual variability depending on the ratio of different bloodsubsets, age, gender, and even the time of day when the bloodsample was collected [82]. This has an important meaning forthe identification of circulating biomarkers from blood sam-ples as inter-variation in both cancerous and control samplesets may alter the miRNA levels within samples.

Subsequently, selection of a miRNA biomarker deregulatedas a result of the glioblastoma and not of any other factorbecomes difficult, and accurate patient information capturingthis data needs to be recorded in public miRNA databases.The presence of other diseases or infections, which is likely inolder patients, could also skew data toward deregulatedmiRNAs as a result of an unrelated pathology leading to thefailure of that miRNA biomarker in further studies. MiRNAexpression and levels are also subject to change followingtherapeutic treatment [83]. MiR-425-5p and 93-5p abundancealters in response to radiochemotherapy in head and necksquamous cell carcinoma patients and could be used as abiomarker to monitor response to treatment. At the same time,this study illustrates the importance of knowing the origin ofsamples used in studies and any treatment undertaken beforecollection of the sample. Ideally, samples for use in biomarkeridentification studies, particularly for diagnostic markers,should be taken before any treatment or surgery to ensureany candidate miRNAs are up-regulated as a result of theglioblastoma and not the response of the tumor to treatment.In addition, this further highlights the need to analyze datasubsets where patterns of miRNA profiles can be extrapolatedfrom patients grouped depending on age, gender, samplingtime, treatment, and other variables.

The methods of isolating and measuring miRNA expres-sion employed in a study can also affect the identification ofmiRNA biomarkers and cause variability between studies. Acomparison of two RNA isolation methods and their effect onserum microarray expression analysis identified differences inmiRNA expression between the two methods. Overall, ahigher expression of miRNAwas observed when using a totalRNA isolation method (guanidine isothiocyanate) comparedto a silica-gel column-based method for isolating small RNAalone [84]. It was concluded in the latter that the high lipidcontent of serum samples affected the isolation of RNA. Thetotal RNA isolation method removes lipids from the samplebefore isolation and therefore provides a better recovery ofRNA [84]. Following isolation, miRNA expression is fre-quently determined using microarray and qRT-PCR methods.The data produced from these methods requires an endoge-nous candidate ubiquitously expressed across all samples fornormalization [85]. For serum and plasma samples, there havebeen no significant endogenous control established for use innormalization [86]. As a result, researchers use a number ofdifferent methods such as mathematical models or syntheticmiRNA spike-ins to normalize their data. While this permitsresearch into circulating miRNA to be performed, theresulting data from different studies is often incomparableand not reproducible contributing to the failure of these bio-markers to reach clinical trials. Standardization of methods forextraction and data analysis would therefore improve thereproducibility of biomarker research and speed the discoveryof a clinically effective biomarker panel.

Mol Neurobiol

There is a need for standardization of isolation and analysistechniques to improve the reliability of candidate miRNAbiomarker data. A large-scale multi-center study into the opti-mal techniques and protocols for miRNA biomarker discoveryfrom a range of sample types including tissues and biofluidswould provide data which could be used to reach a consensuson standard techniques to be used by all researchers. Thiswould subsequently improve the success of miRNA bio-markers for all disease states including glioma. Conducting astudy to identify miRNA biomarkers from biofluids must bestrictly regulated throughout the whole process from patientselection and sample collection through to processing andanalysis. Stringent guidelines to minimize variability in patientcohorts and validate the source of miRNAs, normalizationmethods, and good experimental design should be agreed uponto ensure reproducibility of data and the efficacy of thesemarkers during further studies and clinical trials.

Concluding Remarks

MiRNAs hold the potential to provide sensitive and specificbiomarker panels, not only for glioma but for all cancer types.Their tissue and disease specificity make them ideal candi-dates for diagnostic and prognostic indicators. As high-throughput techniques continue to improve, the current limi-tations of miRNA and biomarker discovery will be overcome.Many of the findings discussed in this review highlight thefirst steps toward the identification of circulating miRNAbiomarkers for glioma diagnosis, prognosis, and grading.Current research in our laboratory corroborates many of thefindings discussed in this review (unpublished data) and high-lights the first steps toward the identification of serum orcerebrospinal fluid miRNA biomarkers for glioma diagnosis,prognosis, and grading. The initial understanding of miRNAinfluence on glioma biology provides the foundation for ef-fective biomarker panels as summarized in Table 3. In turn,these biomarkers can serve as targets for treatment, showingthe evolution of miRNA functional studies into therapeuticstrategies for glioma and ultimately improve the outlook ofindividuals with glioma.

Acknowledgments The authors would like to acknowledge KatherineAshton and Mohit Arora at Royal Preston Hospital for provision ofpatient samples and Brain Tumour North West and the University ofCentral Lancashire for funding.

References

1. Preusser M (2012) Clinically useful biomarkers in neurooncology.memo—Magazine of. Eur Med Oncol 5(3):201–204. doi:10.1007/s12254-012-0030-3

2. Manne U, Srivastava R-G, Srivastava S (2005) Keynote review:recent advances in biomarkers for cancer diagnosis and treatment.Drug Discov Today 10(14):965–976. doi:10.1016/S1359-6446(05)03487-2

3. Lenos K, Grawenda AM, Lodder K, Kuijjer ML, Teunisse AFAS,Repapi E, Grochola LF, Bartel F, Hogendoorn PCW, Wuerl P,Taubert H, Cleton-Jansen A-M, Bond GL, Jochemsen AG (2012)Alternate splicing of the p53 inhibitor HDMX offers a superiorprognostic biomarker than p53 mutation in human cancer. CancerRes 72(16):4074–4084. doi:10.1158/0008-5472.can-12-0215

4. Zougman A, Hutchins GG, Cairns DA, Verghese E, Perry SL, JayneDG, Selby PJ, Banks RE (2013) Retinoic acid-induced protein 3:identification and characterisation of a novel prognostic colon cancerbiomarker. Eur J Cancer 49(2):531–539. doi:10.1016/j.ejca.2012.07.031

5. Bauer TM, El-Rayes BF, Li X, Hammad N, Philip PA, Shields AF,Zalupski MM, Bekaii-Saab T (2013) Carbohydrate antigen 19-9 is aprognostic and predictive biomarker in patients with advanced pan-creatic cancer who receive gemcitabine-containing chemotherapy.Cancer 119(2):285–292. doi:10.1002/cncr.27734

6. Qian X, Li C, Pang B, Xue M, Wang J, Zhou J (2012) Spondin-2(SPON2), a more prostate-cancer-specific diagnostic biomarker.PLoS ONE 7(5):e37225. doi:10.1371/journal.pone.0037225

7. Morrison DH, Rahardja D, King E, Peng Y, Sarode VR (2012)Tumour biomarker expression relative to age and molecular subtypesof invasive breast cancer. Br J Cancer 107(2):382–387

8. Diamandis (2010) Cancer biomarkers: can we turn recent failuresinto success? J Natl Cancer Inst 102(19):1462–1467. doi:10.1093/jnci/djq306

9. Abu-Asab MS, Chaouchi M, Alesci S, Galli S, Laassri M, CheemaAK, Atouf F, VanMeter J, Amri H (2011) Biomarkers in the age ofomics: time for a systems biology approach. OMICS J Integr Biol15(3):105–112

10. Chauhan S, Kumar D, Jaggi M (2009) Mucins in ovarian cancerdiagnosis and therapy. J Ovarian Res 2(1):21

11. Favilla V, Cimino S, Madonia M, Morgia G (2010) New advances inclinical biomarkers in testis cancer. Front Biosci (Elite edition) 2:456–477

12. Lam L, Czerniecki BJ, Fitzpatrick E, Xu S, Schuchter L, Xu X,Zhang H (2013) Interference-free HER2 ECD as a serum biomarkerin breast cancer. J Mol Biomark Diagn 4(151)

13. Wang, Yuan X, Zhou Z, Hu J, Zhang T, Hu S, Luo J, Li X (2011)MicroRNAs might be promising biomarkers of human gliomas.Asian Pac J Cancer Prev 12:833–835

14. Baraniskin A, Kuhnhenn J, Schlegel U, Maghnouj A, Zöllner H,Schmiegel W, Hahn S, Schroers R (2012) Identification ofmicroRNAs in the cerebrospinal fluid as biomarker for thediagnosis of glioma. Neuro-Oncology 14(1):29–33. doi:10.1093/neuonc/nor169

15. Kuratsu J-i, Yoshizato K, Yoshimura T, Leonard EJ, Takeshima H,Ushio Y (1993) Quantitative study of monocyte chemoattractantprotein-1 (MCP-1) in cerebrospinal fluid and cyst fluid from patientswith malignant glioma. J Natl Cancer Inst 85(22):1836–1839. doi:10.1093/jnci/85.22.1836

16. Su X, Chakravarti D, ChoMS, Liu L, Gi YJ, Lin Y-L, LeungML, El-Naggar A, Creighton CJ, Suraokar MB, Wistuba I, Flores ER (2010)TAp63 suppresses metastasis through coordinate regulation of Dicerand miRNAs. Nature 467(7318):986–990. doi:10.1038/nature09459

17. Louis D, Ohgaki H, Wiestler O, Cavenee W, Burger P, Jouvet A,Scheithauer B, Kleihues P (2007) The 2007 WHO classification oftumours of the central nervous system. Acta Neuropathol 114(2):97–109. doi:10.1007/s00401-007-0243-4

18. Yoon DJ, Kwan BH, Chao FC, Nicolaides TP, Phillips JJ, Lam GY,Mason AB, Weiss WA, Kamei DT (2010) Intratumoral therapy ofglioblastoma multiforme using genetically engineered transferrin fordrug delivery. Cancer Res 70(11):4520–4527. doi:10.1158/0008-5472.can-09-4311

Mol Neurobiol

19. TainskyMA (2009) Genomic and proteomic biomarkers for cancer: amultitude of opportunities. Biochim Biophys Acta 1796(2):176–193.doi:10.1016/j.bbcan.2009.04.004

20. Garcia-Bilbao A, Armananzas R, Ispizua Z, Calvo B, Alonso-VaronaA, Inza I, Larranaga P, Lopez-Vivanco G, Suarez-Merino B,Betanzos M (2012) Identification of a biomarker panel for colorectalcancer diagnosis. BMC Cancer 12(1):43

21. Urquidi V, Goodison S, Cai Y, Sun Y, Rosser CJ (2012) A candidatemolecular biomarker panel for the detection of bladder cancer.Cancer Epidemiol Biomark Prev 21(12):2149–2158. doi:10.1158/1055-9965.epi-12-0428

22. Barderas R, Babel I, Díaz-Uriarte R, Moreno V, Suárez A, Bonilla F,Villar-Vázquez R, Capellá G, Casal JI (2012) An optimized predictorpanel for colorectal cancer diagnosis based on the combination oftumor-associated antigens obtained from protein and phage microar-rays. J Proteome 75(15):4647–4655. doi:10.1016/j.jprot.2012.03.004

23. Gabriely G, Yi M, Narayan RS, Niers JM, Wurdinger T, Imitola J,Ligon KL, Kesari S, Esau C, Stephens RM, Tannous BA, KrichevskyAM (2011) Human glioma growth is controlled by microRNA-10b.Cancer Res 71(10):3563–3572. doi:10.1158/0008-5472.can-10-3568

24. Lu J, Getz G, Miska EA, Alvarez-Saavedra E, Lamb J, Peck D,Sweet-Cordero A, Ebert BL, Mak RH, Ferrando AA, Downing JR,Jacks T, Horvitz HR, Golub TR (2005) MicroRNA expression pro-files classify human cancers. Nature 435(7043):834–838. doi:10.1038/nature03702

25. Rosenfeld N, Aharonov R, Meiri E, Rosenwald S, Spector Y,Zepeniuk M, Benjamin H, Shabes N, Tabak S, Levy A, LebanonyD, Goren Y, Silberschein E, Targan N, Ben-Ari A, Gilad S, Sion-Vardy N, Tobar A, Feinmesser M, Kharenko O, Nativ O, Nass D,Perelman M, Yosepovich A, Shalmon B, Polak-Charcon S, FridmanE, Avniel A, Bentwich I, Bentwich Z, Cohen D, Chajut A, Barshack I(2008) MicroRNAs accurately identify cancer tissue origin. NatBiotechnol 26(4):462–469. doi:10.1038/nbt1392

26. Duffy MJ, O’Donovan N, Crown J (2011) Use of molecular markersfor predicting therapy response in cancer patients. Cancer Treat Rev37(2):151–159. doi:10.1016/j.ctrv.2010.07.004

27. Hua D, Mo F, Ding D, Li L, Han X, Zhao N, Foltz G, Lin B, Lan Q,Huang Q (2012) A catalogue of glioblastoma and brain microRNAsidentified by deep sequencing. OMICS J Integr Biol 16(12):690–699

28. Srinivasan S, Patric IRP, Somasundaram K (2011) A ten-microRNAexpression signature predicts survival in glioblastoma. PLoS ONE6(3):e17438. doi:10.1371/journal.pone.0017438

29. Delfino KR, Serão NVL, Southey BR, Rodriguez-Zas SL (2011)Therapy-, gender- and race-specific microRNAmarkers, target genesand networks related to glioblastoma recurrence and survival. CancerGenomics Proteomics 8(4):173–183

30. Kim T-M, Huang W, Park R, Park PJ, Johnson MD (2011) Adevelopmental taxonomy of glioblastoma defined and maintainedby microRNAs. Cancer Res 71(9):3387–3399. doi:10.1158/0008-5472.can-10-4117

31. Li R, GaoK, Luo H,Wang X, Shi Y, Dong Q, LuanW, You Y (2014)Identification of intrinsic subtype-specific prognostic microRNAs inprimary glioblastoma. J Exp Clin Cancer Res 33(9)

32. Lee RC, Feinbaum RL, Ambros V (1993) The C. elegansheterochronic gene lin-4 encodes small RNAs with antisense com-plementarity to lin-14. Cell 75(5):843–854. doi:10.1016/0092-8674(93)90529-Y

33. Graves P, Zeng Y (2012) Biogenesis of mammalian microRNAs: aglobal view. Genomics Proteomics Bioinforma 10(5):239–245. doi:10.1016/j.gpb.2012.06.004

34. Kawahara Y, Megraw M, Kreider E, Iizasa H, Valente L,Hatzigeorgiou AG, Nishikura K (2008) Frequency and fate ofmicroRNA editing in human brain. Nucleic Acids Res 36(16):5270–5280. doi:10.1093/nar/gkn479

35. Yang W, Chendrimada TP, Wang Q, Higuchi M, Seeburg PH,Shiekhattar R, Nishikura K (2006) Modulation of microRNA

processing and expression through RNA editing by ADAR deami-nases. Nat Struct Mol Biol 13(1):13–21. doi:10.1038/nsmb1041

36. Galeano F, Tomaselli S, Locatelli F, Gallo A (2012) A-to-I RNAediting: the “ADAR” side of human cancer. Semin Cell Dev Biol23(3):244–250. doi:10.1016/j.semcdb.2011.09.003

37. Choudhury Y, Tay FC, Lam DH, Sandanaraj E, Tang C, Ang B-T,Wang S (2012) Attenuated adenosine-to-inosine editing ofmicroRNA-376a* promotes invasiveness of glioblastoma cells. JClin Invest 122(11):4059–4076. doi:10.1172/jci62925

38. Yang J-S, Phillips MD, Betel D, Mu P, Ventura A, Siepel AC, ChenKC, Lai EC (2011) Widespread regulatory activity of vertebratemicroRNA* species. RNA 17(2):312–326. doi:10.1261/rna.2537911

39. Chen L, Chen X-R, Zhang R, Li P, Liu Y, Yan K, Jiang X-D (2013)MicroRNA-107 inhibits glioma cell migration and invasion by mod-ulating Notch2 expression. J Neurooncol 112(1):59–66. doi:10.1007/s11060-012-1037-7

40. Loftus JC, Ross JT, Paquette KM, Paulino VM, Nasser S, Yang Z,Kloss J, Kim S, Berens ME, Tran NL (2012) miRNA expressionprofiling in migrating glioblastoma cells: regulation of cell migrationand invasion by miR-23b via targeting of Pyk2. PLoS ONE 7(6):e39818–e39818. doi:10.1371/journal.pone.0039818

41. Tan X, Wang S, Zhu L, Wu C, Yin B, Zhao J, Yuan J, Qiang B, PengX (2012) cAMP response element-binding protein promotesgliomagenesis by modulating the expression of oncogenicmicroRNA-23a. Proc Natl Acad Sci 109(39):15805–15810. doi:10.1073/pnas.1207787109

42. Song L, Liu L,Wu Z, Li Y, Ying Z, Lin C,Wu J, HuB, Cheng S-Y, LiM, Li J (2012) TGF-β induces miR-182 to sustain NF-κB activationin glioma subsets. J Clin Invest 122(10):3563–3578. doi:10.1172/jci62339

43. Karin M, Cao Y, Greten FR, Li Z-W (2002) NF-[kappa]B in cancer:from innocent bystander tomajor culprit. Nat RevCancer 2(4):301–310

44. Jiang L, Lin C, Song L,Wu J, Chen B, Ying Z, Fang L, Yan X, HeM,Li J, Li M (2012) MicroRNA-30e* promotes human glioma cellinvasiveness in an orthotopic xenotransplantation model bydisrupting the NF-κB/IκBα negative feedback loop. J Clin Invest122(1):33–47. doi:10.1172/jci58849

45. Godlewski J, Nowicki MO, Bronisz A, Nuovo G, Palatini J, De LayM, Van Brocklyn J, Ostrowski MC, Chiocca EA, Lawler SE (2010)MicroRNA-451 regulates LKB1/AMPK signaling and allows adap-tation to metabolic stress in glioma cells. Mol Cell 37(5):620–632.doi:10.1016/j.molcel.2010.02.018

46. Yang, Wang C, Chen X, Chen S, Zhang Y, Zhi F, Wang J, Li L, ZhouX, Li N, Pan H, Zhang J, Zen K, Zhang C-Y, Zhang C (2013)Identification of seven serum microRNAs from a genome-wide se-rum microRNA expression profile as potential noninvasive bio-markers for malignant astrocytomas. Int J Cancer 132(1):116–127.doi:10.1002/ijc.27657

47. Trajkovic K, Hsu C, Chiantia S, Rajendran L, Wenzel D, Wieland F,Schwille P, Brügger B, Simons M (2008) Ceramide triggers buddingof exosome vesicles into multivesicular endosomes. Science319(5867):1244–1247. doi:10.1126/science.1153124

48. Ostrowski M, Carmo NB, Krumeich S, Fanget I, Raposo G, SavinaA,Moita CF, Schauer K, Hume AN, Freitas RP, Goud B, Benaroch P,Hacohen N, Fukuda M, Desnos C, Seabra MC, Darchen F,Amigorena S,Moita LF, Thery C (2010) Rab27a and Rab27b controldifferent steps of the exosome secretion pathway. Nat Cell Biol 12(1):19–30. doi:10.1038/ncb2000

49. Bobrie A, Krumeich S, Reyal F, Recchi C, Moita LF, Seabra MC,Ostrowski M, Théry C (2012) Rab27a supports exosome-dependentand -independent mechanisms that modify the tumor microenviron-ment and can promote tumor progression. Cancer Res 72(19):4920–4930. doi:10.1158/0008-5472.can-12-0925

50. Chen X, Liang H, Zhang J, Zen K, Zhang C-Y (2012) SecretedmicroRNAs: a new form of intercellular communication. TrendsCell Biol 22(3):125–132. doi:10.1016/j.tcb.2011.12.001

Mol Neurobiol

51. Skog J, Wurdinger T, van Rijn S, Meijer DH, Gainche L, Curry WT,Carter BS, Krichevsky AM, Breakefield XO (2008) Glioblastomamicrovesicles transport RNA and proteins that promote tumourgrowth and provide diagnostic biomarkers. Nat Cell Biol 10(12):1470–1476. doi:10.1038/ncb1800

52. Li CCY, Eaton SA, Young PE, Lee M, Shuttleworth R, HumphreysDT, GrauGE, Combes V, BebawyM, Gong J, Brammah S, BucklandME, Suter CM (2013) Glioma microvesicles carry selectively pack-aged coding and non-coding RNAs which alter gene expression inrecipient cells. RNA Biol 10(8):1333–1344

53. Chen C, Ridzon DA, Broomer AJ, Zhou Z, Lee DH, Nguyen JT,Barbisin M, Xu NL, Mahuvakar VR, Andersen MR, Lao KQ, LivakKJ, Guegler KJ (2005) Real-time quantification of microRNAs bystem-loop RT-PCR. Nucleic Acids Res 33(20):e179. doi:10.1093/nar/gni178

54. Sheinerman KS, Umansky SR (2013) Circulating cell-freemicroRNA as biomarkers for screening, diagnosis and monitoringof neurodegenerative diseases and other neurologic pathologies.Front Cell Neurosci 7:150. doi:10.3389/fncel.2013.00150

55. Wang Q, Li P, Li A, JiangW,Wang H, Wang J, Xie K (2012) Plasmaspecific miRNAs as predictive biomarkers for diagnosis and progno-sis of glioma. J Exp Clin Cancer Res 31(97):1–10

56. de Bont JM, den Boer ML, Reddingius RE, Jansen J, Passier M, vanSchaik RH, Kros JM, Sillevis Smitt PA, Luider TH, Pieters R (2006)Identification of apolipoprotein A-II in cerebrospinal fluid of pediat-ric brain tumor patients by protein expression profiling. Clin Chem52(8):1501–1509. doi:10.1373/clinchem.2006.069294

57. Niclou SP, Fack F, Rajcevic U (2010) Glioma proteomics: status andperspectives. J Proteome 73(10):1823–1838. doi:10.1016/j.jprot.2010.03.007

58. Baraniskin A, Kuhnhenn J, Schlegel U, Maghnouj A, Zollner H,Schmiegel W, Hahn S, Schroers R (2011) Identification ofmicroRNAs in the cerebrospinal fluid as biomarker for the diagnosisof glioma. Neuro-Oncology 117(11):3140–3146. doi:10.1093/neuonc/nor169

59. Teplyuk NM,Mollenhauer B, Gabriely G, Giese A, Kim E, SmolskyM,KimRY, SariaMG, Pastorino S, Kesari S, KrichevskyAM (2012)MicroRNAs in cerebrospinal fluid identify glioblastoma and meta-static brain cancers and reflect disease activity. Neuro-Oncology14(6):689–700. doi:10.1093/neuonc/nos074

60. Gabriely G,Wurdinger T, Kesari S, Esau CC, Burchard J, Linsley PS,Krichevsky AM (2008) MicroRNA 21 promotes glioma invasion bytargeting matrix metalloproteinase regulators. Mol Cell Biol 28(17):5369–5380. doi:10.1128/mcb.00479-08

61. Xia H, QiY, Ng SS, Chen X, Chen S, FangM, Li D, ZhaoY, Ge R, LiG, Chen Y, He M-L, H-f K, Lai L, Lin MC (2009) MicroRNA-15bregulates cell cycle progression by targeting cyclins in glioma cells.Biochem Biophys Res Commun 380(2):205–210. doi:10.1016/j.bbrc.2008.12.169

62. Peng B, Hu S, Jun Q, Luo D, Zhang X, Zhao H, Li D (2013)MicroRNA-200b targets CREB1 and suppresses cell growth in hu-man malignant glioma. Mol Cell Biochem 379(1–2):51–58. doi:10.1007/s11010-013-1626-6

63. Carden CP, Sarker D, Postel-Vinay S, Yap TA, Attard G, Banerji U,Garrett MD, Thomas GV, Workman P, Kaye SB, de Bono JS (2010)Can molecular biomarker-based patient selection in phase I trialsaccelerate anticancer drug development? Drug Discov Today 15(3–4):88–97. doi:10.1016/j.drudis.2009.11.006

64. ZhangW, Zhang J, Hoadley K, Kushwaha D, Ramakrishnan V, Li S,Kang C, You Y, Jiang C, Song SW, Jiang T, Chen CC (2012) miR-181d: a predictive glioblastoma biomarker that downregulatesMGMT expression. Neuro-Oncology 14(6):712–719. doi:10.1093/neuonc/nos089

65. Shi L, Chen J, Yang J, Pan T, Zhang S, Wang Z (2010) MiR-21protected human glioblastoma U87MG cells from chemotherapeuticdrug temozolomide induced apoptosis by decreasing Bax/Bcl-2 ratio

and caspase-3 activity. Brain Res 1352:255–264. doi:10.1016/j.brainres.2010.07.009

66. WONG STS, X-Q ZHANG, ZHUANG JT-F, H-L CHAN, C-H LI,LEUNG GKK (2012) MicroRNA-21 inhibition enhances in vitrochemosensitivity of temozolomide-resistant glioblastoma cells.Anticancer Res 32(7):2835–2841

67. GwakH-S, Kim TH, JoGH, KimY-J, KwakH-J, Kim JH, Yin J, YooH, Lee SH, Park JB (2012) Silencing of microRNA-21 confers radio-sensitivity through inhibition of the PI3K/AKT pathway and enhanc-ing autophagy in malignant glioma cell lines. PLoS ONE 7(10):e47449. doi:10.1371/journal.pone.0047449

68. Quintavalle C, Donnarumma E, Iaboni M, Roscigno G, Garofalo M,Romano G, Fiore D, De Marinis P, Croce CM, Condorelli G (2012)Effect of miR-21 and miR-30b/c on TRAIL-induced apoptosis inglioma cells. Oncogene. doi:10.1038/onc.2012.410

69. Chan LW,Moses MA, Goley E, Sproull M, Muanza T, Coleman CN,Figg WD, Albert PS, Ménard C, Camphausen K (2004) UrinaryVEGF and MMP levels as predictive markers of 1-yearprogression-free survival in cancer patients treated with radiationtherapy: a longitudinal study of protein kinetics throughout tumorprogression and therapy. J Clin Oncol 22(3):499–506. doi:10.1200/jco.2004.07.022

70. Vinther J, Rukov J, Shomron N (2012) MicroRNAs and their antag-onists as novel therapeutics. In: Erdmann VA, Barciszewski J (eds)From nucleic acids sequences to molecular medicine. RNAtechnologies. Springer, Berlin, pp 503–523. doi:10.1007/978-3-642-27426-8_20

71. Nana-Sinkam SP, Croce CM (2013) Clinical applications formicroRNAs in cancer. Clin Pharmacol Ther 93(1):98–104

72. Bouchie A (2013) First microRNA mimic enters clinic. NatBiotechnol 31(7):577–577. doi:10.1038/nbt0713-577

73. Melo S, Villanueva A, Moutinho C, Davalos V, Spizzo R, Ivan C,Rossi S, Setien F, Casanovas O, Simo-Riudalbas L, Carmona J,Carrere J, Vidal A, Aytes A, Puertas S, Ropero S, Kalluri R, CroceCM, Calin GA, Esteller M (2011) Small molecule enoxacin is acancer-specific growth inhibitor that acts by enhancing TAR RNA-binding protein 2-mediated microRNA processing. Proc Natl AcadSci 108(11):4394–4399. doi:10.1073/pnas.1014720108

74. Git A, Dvinge H, Salmon-Divon M, Osborne M, Kutter C, HadfieldJ, Bertone P, Caldas C (2010) Systematic comparison of microarrayprofiling, real-time PCR, and next-generation sequencing technolo-gies for measuring differential microRNA expression. RNA 16(5):991–1006. doi:10.1261/rna.1947110

75. McDonald JS, Milosevic D, Reddi HV, Grebe SK, Algeciras-Schimnich A (2011) Analysis of circulating microRNA:preanalytical and analytical challenges. Clin Chem 57(6):833–840.doi:10.1373/clinchem.2010.157198

76. Qureshi R, Sacan A (2013) A novel method for the normalization ofmicroRNA RT-PCR data. BMC Med Genet 6(1):14–17

77. Duttagupta R, Jiang R, Gollub J, Getts RC, Jones KW (2011) Impactof cellular miRNAs on circulating miRNA biomarker signatures.PLoS ONE 6(6):e20769. doi:10.1371/journal.pone.0020769

78. Zampetaki A, Mayr M (2012) Analytical challenges and technicallimitations in assessing circulating MiRNAs. Thromb Haemost108(10):592–598. doi:10.1160/th12-02-0097

79. Noerholm M, Balaj L, Limperg T, Salehi A, Zhu L, Hochberg F,Breakefield X, Carter B, Skog J (2012) RNA expression patterns inserum microvesicles from patients with glioblastoma multiforme andcontrols. BMC Cancer 12(22):2–11

80. Pritchard CC, Kroh E, Wood B, Arroyo JD, Dougherty KJ, MiyajiMM, Tait JF, Tewari M (2012) Blood cell origin of circulatingmicroRNAs: a cautionary note for cancer biomarker studies. CancerPrev Res 5(3):492–497. doi:10.1158/1940-6207.capr-11-0370

81. Mathers JC, Strathdee G, Relton CL (2010) Induction of epigeneticalterations by dietary and other environmental factors. AdvGenet 71:3–39

Mol Neurobiol

82. Whitney AR, Diehn M, Popper SJ, Alizadeh AA, Boldrick JC,Relman DA, Brown PO (2003) Individuality and variation in geneexpression patterns in human blood. Proc Natl Acad Sci 100(4):1896–1901. doi:10.1073/pnas.252784499

83. Summerer I, Niyazi M, Unger K, Pitea A, Zangen V, Hess J,Atkinson MJ, Belka C, Moertl S, Zitzelsberger H (2013) Changesin circulating microRNAs after radiochemotherapy in head and neckcancer patients. Radiat Oncol 8(296)

84. Podolska A, Kaczkowski B, Litman T, Fredholm M, Cirera S (2011)How the RNA isolation method can affect microRNA microarrayresults. Acta Biochim Pol 58(4):535–540

85. Peltier HJ, Latham GJ (2008) Normalization of microRNA expres-sion levels in quantitative RT-PCR assays: identification of suitablereference RNA targets in normal and cancerous human solid tissues.RNA 14(5):844–852. doi:10.1261/rna.939908

86. Mitchell PS, Parkin RK, Kroh EM, Fritz BR, Wyman SK,Pogosova-Agadjanyan EL, Peterson A, Noteboom J, O'BriantKC, Allen A, Lin DW, Urban N, Drescher CW, Knudsen BS,Stirewalt DL, Gentleman R, Vessella RL, Nelson PS, MartinDB, Tewari M (2008) Circulating microRNAs as stable blood-based markers for cancer detection. Proc Natl Acad Sci105(30):10513–10518. doi:10.1073/pnas.0804549105

Mol Neurobiol