Search tips
Search criteria 


Logo of neuroncolAboutAuthor GuidelinesEditorial BoardNeuro-Oncology
Neuro Oncol. 2016 May; 18(5): 656–666.
Published online 2015 September 15. doi:  10.1093/neuonc/nov196
PMCID: PMC4827035

Coordination of self-renewal in glioblastoma by integration of adhesion and microRNA signaling



Cancer stem cells (CSCs) provide an additional layer of complexity for tumor models and targets for therapeutic development. The balance between CSC self-renewal and differentiation is driven by niche components including adhesion, which is a hallmark of stemness. While studies have demonstrated that the reduction of adhesion molecules, such as integrins and junctional adhesion molecule-A (JAM-A), decreases CSC maintenance. The molecular circuitry underlying these interactions has yet to be resolved.


MicroRNA screening predicted that microRNA-145 (miR-145) would bind to JAM-A. JAM-A overexpression in CSCs was evaluated both in vitro (proliferation and self-renewal) and in vivo (intracranial tumor initiation). miR-145 introduction into CSCs was similarly assessed in vitro. Additionally, The Cancer Genome Atlas dataset was evaluated for expression levels of miR-145 and overall survival of the different molecular groups.


Using patient-derived glioblastoma CSCs, we confirmed that JAM-A is suppressed by miR-145. CSCs expressed low levels of miR-145, and its introduction decreased self-renewal through reductions in AKT signaling and stem cell marker (SOX2, OCT4, and NANOG) expression; JAM-A overexpression rescued these effects. These findings were predictive of patient survival, with a JAM-A/miR-145 signature robustly predicting poor patient prognosis.


Our results link CSC-specific niche signaling to a microRNA regulatory network that is altered in glioblastoma and can be targeted to attenuate CSC self-renewal.

Keywords: cancer stem cell, glioblastoma, JAM-A, miR-145

Cellular heterogeneity is appreciated as a hallmark of advanced tumors, and this phenotype has recently been appreciated as a contributing factor in the complexity of cancer. Functional studies have sought to determine how distinct cell populations contribute to tumor growth and therapeutic resistance. These efforts have led to the identification of a population of tumor cells with enhanced self-renewal and tumor-initiation capacities that possess stem cell-like features termed cancer stem cells (CSCs).1 CSCs are a dynamic population, maintained in discrete anatomical niches, that promote their self-renewal, tumor maintenance, and increased resistance to conventional therapies.2 Glioblastoma (GBM), the most prevalent malignant primary brain tumor, contains CSCs37 and has been a prototypic tumor for the study of CSC biology. Despite aggressive treatment consisting of maximal safe surgical resection, radiation, and chemotherapy, 5-year overall survival for GBM patients is <10%.8,9 Identification and subsequent targeting of mechanisms responsible for the maintenance of CSCs in combination with current GBM treatments may have a synergistic therapeutic effect and therefore improve patient prognosis.

To fulfill the eventual goal of developing CSC-targeted therapies, the identification of CSC-specific regulatory mechanisms is required. Based on the importance of niche interactions to CSC maintenance, interrogating the signaling mechanisms present within the niche remains a priority. In a manner similar to the dependence of neural progenitor cells (NPCs) on extracellular interactions with their microenvironment,10,11 the interaction of CSCs with their niche via adhesion proteins is critical for their maintenance in GBM.1214 While adhesion molecules involved in CSC-niche interactions that drive CSC maintenance and therapeutic resistance in GBM have been identified,1215 the molecular circuitry responsible for regulating adhesion molecules and how they are integrated into larger signaling networks has yet to be determined. Multiple phenotypes in GBM, including tumor suppression,16 therapeutic resistance,17 and self-renewal16,1820 have been linked to miRNAs. However, the interaction between miRNAs and niche-adhesion molecules remains largely unexplored. Using junctional adhesion molecule A (JAM-A; a CSC-specific adhesion protein), as a paradigm, we sought to identify the miRNA regulatory circuitry linking niche adhesion molecules to a larger signaling network. Through a screen for miRNAs that bind to JAM-A and are downregulated in GBM, we identified miR-145 as a negative regulator of JAM-A-mediated CSC maintenance.

Materials and Methods

Cancer Stem Cell Derivation, Culture, and Analysis

Established GBM xenografts representing the classical (T4121), mesenchymal (T387), and proneural (T3691) subtypes were used, as previously reported.12,2123 GBM cells were dissociated from established xenografts under Cleveland Clinic-approved Institutional Animal Care and Use Committee protocols. Xenografts were passaged in immune-deficient athymic nude or NSG mice (The Jackson Laboratory) for maintenance of tumor heterogeneity. Briefly, 6-week-old female mice were injected subcutaneously with freshly dissociated human GBM cells, and tumor cells were allowed to grow until volume exceeded 5% of the animal's body weight, after which the mice were euthanized. Xenografted tumors were dissected and mechanically dissociated using papain dissociation kits (Worthington Biochemical Corporation), and cells were cultured overnight in neurobasal medium (Life Technologies) supplemented with B27 (Life Technologies), 1% penicillin/streptomycin (Life Technologies), sodium pyruvate 1:100, L-glutamine 1:100, EGF (R&D Systems, 20 ng/mL), and FGF-2 (R&D Systems 20 ng/mL). CSCs were enriched using CD133/2 magnetic beads (Miltenyi Biotech) and cultured in supplemented neurobasal medium. Non-stem cells (non-CSCs) were obtained in parallel and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Sigma) and 1% penicillin/streptomycin. Both tumor cell populations were cultured in the appropriate complete medium until the day they were used. However, only low-passage (<5) cells were used for experiments to prevent cellular drift.

In Vitro Functional Analysis: Tumorsphere Formation and Cell Proliferation

Cell proliferation experiments were conducted by plating cells of interest at a density of 1000 cells/well in a 96-well plate in triplicate. Cell number was measured every other day and normalized to the initial reading at day zero using the CellTiter-Glo assay kit (Promega). For tumorsphere formation experiments, cells were sorted using a flow cytometer (FACSAria II) into 96-well plates at a density of 1, 5, 10, and 20 live cells per well (24 wells for each density). Cells were maintained for 10 days before sphere formation was evaluated. Spheres larger than 10 cells in diameter were considered for analysis. Reported numbers represent either number of cells per well or stem cell frequency calculated using the Walter and Eliza Hall Institute Bioinformatics Division ELDA analyzer (

Immunoblotting Analysis

Cell populations were lysed using RIPA lysis buffer (containing PMSF, protease inhibitor cocktail, and sodium orthovanadate; Santa Cruz BiotechnologyA), and protein concentrations were calculated using a BCA protein assay (Pierce Biotechnology). After denaturation with Laemmli buffer (BioRad Laboratories), 10 µg of total protein were loaded on 12% polyacrylamide SDS-PAGE gels, transferred to polyvinyl difluoride membranes (Millipore) and probed using the following antibodies: JAM-A (B&D Biosciences, 1:1000), SOX2 (R&D Systems, 1:1000), AKT (Cell Signaling, 1:2000), and p-AKT (Cell Signaling, 1:2000); β-Actin (Santa Cruz Biotechnology, 1:2000) was used as a loading control. Species-specific horseradish peroxidase-conjugated secondary antibodies were used for detection (Invitrogen, 1:5000). Membranes were developed using ECL-2 reagent (Pierce Biotechnology).

miRWalk Database and Sequence Alignment

Using the predicted gene-miRNA interaction search, a list of candidate miRNAs was generated. Results were validated by comparison with other databases, and only those miRNAs present in an additional 3 databases were included in Fig. 2. The sequence alignment between miR145 and JAM-A was generated using miRanda (25;

Fig. 2.
Junctional adhesion molecule-A (JAM-A) is a target of miR-145, which is downregulated in cancer stem cells (CSCs). Schematic of miRNAs predicted to bind to JAM-A generated with miRwalk and miRNAs reported to be downregulated in glioblastoma (A) illustrating ...

Lentiviral shRNA and Overexpression Construct Preparation

Lentiviral constructs were prepared according to modified protocols from Tronolab. In short, using calcium phosphate precipitation, 293FT cells were co-transfected with the packaging vectors psPAX2 and pMD2.G (Addgene) and lentiviral vectors directing the expression of (i) MISSION shRNA (Sigma) specific to SOX2: (TRCN0000003252 (KD1) and TRCN0000003253 (KD2) or a nontargeting control (NT) shRNA (SHC002) and (ii) overexpression of JAM-A: accession number BC001533 (LV152204 (Applied Biological Materials) or control vector (LV590) to produce virus. Media on the 293FT cell cultures were changed 18 hours after transfection, and viral supernatants were collected 12, 24, and 36 hours later and concentrated using polyethylene glycol precipitation for immediate use or stored at −80°C for future use.

In Vivo Intracranial Injections

For in vivo tumor formation, JAM-A or control vector-containing live CSCs was transplanted into the frontal lobe of NOD scid gamma (NSG) mice at 100 or 1000 cells per mouse (n = 8). Mice were monitored daily and euthanized upon the development of neurological signs.

miRNA-145 Introduction

Approximately 2 × 106 cells were transfected with 20 pmol of miR-145 or NT control mimics (Dharmacon) utilizing an Amaxa Nucleofector II and the Mouse Neural Stem Cell Nucleofector Kit (Lonza) as previously described.26 The transfected cells were then collected after 3 days and used for downstream analyses including immunoblotting and quantitative PCR.

Luciferase Expression

Cells containing NT control or miR-145 mimics were used for transfection with luciferase constructs. A total of 20 000 cells per condition were plated in triplicate in 96-well plates pretreated with Geltrex (Life Technologies), which was used as an adherence substrate. The next day, cells were transfected with luciferase only, luciferase + 3′UTR, or luciferase + 3′UTR without seed sequence (ACAATGGACCTTTTGAACTGGAA) constructs using Lipofectamine 2000 (Life Technologies) with 0.6 µg of DNA per reaction in Opti-Mem (Lerner Research Institute Media Core) medium. After 6 hours at 37°C, cells were washed, and luciferase levels were measured 48 hours later using Dual-Glo Luciferase Assay (Promega) per the manufacturer's instructions.

Real-Time Reverse Transcription Polymerase Chain Reaction (qRT-PCR)

RNA from cells of interest was extracted using TRIzol (Life Technologies), and cDNA was synthesized using the Superscript III kit (Invitrogen). qPCR reactions were performed using an ABI 7900HT system having SYBR-Green Mastermix (SA Biosciences). For qPCR analysis, the threshold cycle (CT) values for each gene were normalized to expression levels of β-Actin. Dissociation curves were evaluated for primer fidelity, and only threshold cycles below 35 cycles were reported.

The following primers (Integrated DNA Technologies) were used:

  • β-Actin: Forward 5′-AGAAAATCTGGCACCACACC-3′

For microRNA analysis, we used the Taqman MicroRNA Cells-to-Ct Kit (Applied Biosystems) per the manufacturer's instructions. miRNA-145 and control U6 snRNA 5x and 20x primers (PN4427975 (Applied Biosystems) were used with this kit.

Patient Database Bioinformatics

Gene and miRNA expression data were obtained from The Cancer Genome Atlas (TCGA; for patients with GBM.27 Patients were divided into high and low groups based on mean ± one standard deviation, respectively. Kaplan-Meier survival curves were generated comparing these 2 groups via log-rank test. microRNA data were combined with expression data for JAM-A, NANOG, or SOX2 by subtracting the expression data from the miR-145 data for each patient, and the analysis described above was repeated. Patients were also divided based on their molecular subtype and compared with each other using 1-way ANOVA. Combined microRNA/gene expression analyses were also performed for each of the molecular subtypes.

Statistical Analysis

Reported values are mean values ± standard error of the mean from studies performed at least in triplicate. Unless otherwise stated, 1-way ANOVA was used to calculate statistical significance, with P values detailed in the text and figure legends.


JAM-A Gain of Function Increases Proliferation, Self-renewal, and Tumor Initiation In Vivo

We previously demonstrated that JAM-A was essential for CSC maintenance.15 To assess the sufficiency of JAM-A for driving CSC malignancy and aggressiveness, we evaluated the effect of JAM-A overexpression. JAM-A was overexpressed in CSCs using a lentiviral vector (JAM-A vector) to achieve stable transfection and led to an increase in proliferation compared with CSCs treated with control vector (Fig. 1A). Similarly, self-renewal was elevated in CSCs treated with the JAM-A vector as assessed by in vitro limiting dilution analysis (Fig. 1C); the stem cell frequencies generated showed an increase from 1:7, 1:11, and 1:17 to 1:3, 1:7, and 1:9 in JAM-A overexpressing CSCs compared with control vector, respectively, in the specimens analyzed. Likewise, this increase in proliferation was also observed in non-CSCs when JAM-A was overexpressed using the same lentiviral system (Supplementary material, Fig. S1B). We evaluated several key signaling nodes in CSCs and found that JAM-A overexpression was associated with an increase in p-AKT levels and SOX2 expression (Fig. 1B and Supplementary material, Fig. S1A). Next, we evaluated whether differences in tumor initiation and growth could be seen in vivo by intracranially transplanting CSCs containing control or JAM-A vectors. The median survival of mice injected with 1000 cells per mouse was reduced from 42 days in the blank vector to 34 days in the JAM-A vector group (Fig. 1D). An equally significant decrease in median survival was seen when 100 cells per mouse were intracranially transplanted (62 days in the blank vector vs 47 days in the JAM-A vector group [a reduction of 24%]; data not shown). These data verify the importance of JAM-A in CSC self-renewal and demonstrate that JAM-A overexpression drives CSC marker expression, proliferation, and self-renewal in vitro and tumor initiation in vivo.

Fig. 1.
Junctional adhesion molecule-A (JAM-A) gain-of-function increases proliferation, self-renewal, and tumor initiation in vivo. JAM-A overexpression in cancer stem cells (CSCs) increased proliferation. (A) Immunoblots demonstrate that CSCs overexpressing ...

JAM-A Is a Target of miR-145, Which is Downregulated in Cancer Stem Cells

CSCs do not operate alone but rather respond to and interact with components of their tumor microenvironment. As adhesion molecules are an essential part of this niche, we investigated the upstream regulatory mechanism behind JAM-A expression. To couple adhesion to a larger signaling network, we identified miRNAs that are predicted to bind to JAM-A using the miRWalk database.28 We compared this list to miRNAs reported to be downregulated in GBM29,30 and found miR-145 as a potential target (Fig. 2A). We further validated these results in 2 independent GBM cohorts and found that miR-145 expression was decreased in GBM compared with neural progenitor cells (Supplementary material, Fig. S2A) and nonneoplastic brain tissue (Supplementary material, Fig. S2B). miR-145 has been previously associated with tumor suppressor functions in GBM via suppression of neural precursor cell expressed developmentally downregulated 9 (NEDD9), a scaffolding protein involved in invasion.31 Sequence alignment suggested that miR-145 binds to the 3′UTR region of the JAM-A mRNA to block its translation or promote its degradation (Fig. 2B). Therefore, we used the 3′UTR region of JAM-A in luciferase assays to confirm that JAM-A is a direct target of miR-145. In CSCs where we introduced a control nontargeting microRNA (NT mimics), we observed that both the luciferase control and the luciferase construct with JAM-A 3′UTR were expressed at similar levels (Supplementary material, Fig. S3A). However, when miR-145 was introduced (miR-145 mimics), we observed a decrease in expression only in the luciferase with JAM-A 3′UTR construct (Supplementary material, Fig. S3A). To further confirm that the difference observed was due to miR-145 binding to the putative binding site in the JAM-A 3′UTR, we also introduced constructs without the seed-binding region (mut) in CSCs. Luciferase levels were comparable in the mut and complete 3′UTR constructs when NT mimics were introduced in both specimens analyzed (Fig. 2C and E). However, upon miR-145 introduction, only the complete 3′UTR evidenced a decrease in luciferase, while the construct lacking the binding region to miR-145 showed no change (Fig. 2C and E). To assess whether the endogenous expression levels of miR-145 were lower in CSCs compared with their non-CSC counterparts, we evaluated expression levels in freshly dissociated xenograft tumors after enrichment for CSCs using CD133 as a surface marker. In all specimens analyzed, miR-145 levels were significantly lower in the CSC population (Fig. 2D and F and Supplementary material, Fig. S3B) compared with the non-CSC population. As expected, JAM-A expression followed the opposite pattern (between 2-fold- and 4-fold increased expression in the CSCs). As a control, levels of SOX2 and OLIG2 were measured, and results confirmed between a 2-fold and 8-fold increase in expression of these CSC markers in CSCs compared with non-CSCs. Similar expression differences were observed when CSCs were compared with unenriched cells from a freshly dissociated tumor (Supplementary material, Fig. S3C).

miR-145 Introduction Downregulates JAM-A and Compromises Self-renewal

We next sought to evaluate the effect of miR-145 gain of function in CSCs. We confirmed that JAM-A mRNA and protein levels were downregulated in CSCs after miR-145 introduction compared with cells containing NT mimics (Fig. 3A and B and Supplementary material, Fig. S4A). Paralleling the results of JAM-A overexpression, the decrease in JAM-A was associated with a decrease in p-AKT and SOX2 (Fig. 3B and Supplementary material, Fig. S4A). We analyzed whether the repression of these proteins had a functional consequence in CSCs using in vitro self-renewal assays. Indeed, introduction of miR-145 decreased the stem cell frequencies as well as the number of spheres formed (Fig. 3C and Supplementary material, Fig. S4B) compared with the NT mimics group. Moreover, miR-145 has been previously reported to regulate pluripotency factors.32 Upon the introduction of miR-145 into CSCs, we found that the levels of SOX2, OCT4, and NANOG were reduced (Fig. 3D), suggesting a role for miR-145 in regulating core self-renewal genes. In addition, we also assessed whether SOX2 levels affected the expression levels of miR-145, as previously hypothesized,33 by knocking down Sox2 using 2 independent shRNA constructs (Supplementary material, Fig. S5A). We measured miR-145 expression levels after Sox2 knockdown and found a reduction, not an induction as previously predicted, in both KD groups compared with the control group (Supplementary material, Fig. S5B). Taken together, these data suggest that lower levels of miR-145 are critical for JAM-A expression (and are not dependent on SOX2 levels) and that miR-145 regulates CSC self-renewal.

Fig. 3.
miR-145 introduction downregulates junctional adhesion molecule-A (JAM-A) and compromises self-renewal. Restoration of miR-145 in cancer stem cells (CSCs) reduces mRNA (A) and protein levels (B) of JAM-A together with protein levels of SOX2 and p-AKT ...

JAM-A Overexpression Rescues CSC Self-Renewal After miR-145 Introduction Through a Double Negative-feedback Mechanism

To confirm that the effects on self-renewal induced by miR-145 were dependent on JAM-A, we analyzed the behavior of JAM-A-overexpressing cells after treatment with NT and miR-145 mimics. We observed that JAM-A protein levels were reduced upon miR-145 introduction but were rescued by JAM-A overexpression (Fig. 4A and B). Importantly, SOX2 protein levels correlated with JAM-A expression; thus, JAM-A overexpression also rescued SOX2 in both specimens analyzed (Fig. 4A and B). Functionally, as assessed by in vitro-limiting dilution analysis, we also demonstrated that the miR-145-induced decrease in self-renewal was rescued by JAM-A overexpression (Fig. 4C and D). Additionally, we observed a reduction in miR-145 levels in CSCs treated with JAM-A vector compared with control vector (Fig. 5A), suggesting a double feedback mechanism. JAM-A overexpression led to an increase in p-AKT, which is key for CSC maintenance34,35 and has been shown to regulate miR-145.36 Thus, we treated CSCs with a PI3K inhibitor (LY294002) or an AKT inhibitor (MK226) and measured the levels of miR-145. Indeed, miR-145 levels were 2–3-fold higher in inhibitor-treated cells compared with the DMSO group (Fig. 5B), confirming a 2-way regulation system in this pathway. The data described above lead to a regulatory axis whereby: (i) JAM-A is preferentially expressed in CSCs, and miR-145 (which regulates JAM-A) is downregulated in the same population; (ii) overexpression of JAM-A is associated with elevated p-AKT levels and decreased miR-145; and (iii) miR-145 also regulates self-renewal genes such as SOX2, OCT4, and NANOG (Fig. 5C). In CSCs, increased levels of JAM-A can block the normal inhibition of stem cell genes by miR-145, thereby indirectly generating an increase in self-renewal and promoting malignancy and tumorigenesis.

Fig. 4.
Junctional adhesion molecule-A) JAM-A overexpression rescues cancer stem cell (CSC) self-renewal after miR-145 introduction. miR-145 introduction into CSCs overexpressing JAM-A does not decrease JAM-A or SOX2 protein levels (A and B). Limiting dilution ...
Fig. 5.
Junctional adhesion molecule-A (JAM-A)/miR-145 double-feedback mechanism model. miR-145 expression levels are downregulated with JAM-A overexpression (A) and upregulated in response to both PI3K (LY) and AKT (MK226) inhibitors (B) Schematic of the proposed ...

Lower miR-145 Expression Is Associated With Decreased Patient Survival

To interrogate the clinical relevance of miR-145 and JAM-A in patient prognosis, we evaluated TCGA dataset27 and found that miR-145 levels were predictive of overall GBM patient survival, with patients with lower levels of miR-145 having significantly poorer prognosis (Fig. 6A). When miR-145 levels were combined with JAM-A expression from the same patients, the difference in survival was enhanced (Fig. 6B), with high JAM-A/low miR-145 correlating with much poorer prognosis. This difference was also observed when miR-145 was combined with NANOG (Fig. 6C) but not with SOX2 (Fig 6D). The median survival was the lowest when low miR-145 was combined with JAM-A (Supplementary material, Table S1), and median survival in general was lower in the groups with low miR-145 levels. This further demonstrates that JAM-A and stem cell genes combined with miR-145 levels can predict survival in GBM patient datasets. Similarly, we evaluated the levels of miR-145 in the different subtypes of GBM in the TCGA dataset and found that the proneural subtype had the lowest expression (Supplementary material, Fig. S6A). Interestingly, a similar analysis evaluating the expression of JAM-A also showed lower expression in the proneural group (Supplementary material, Fig. S6A), suggesting there could be additional components in the JAM-A/miR-145 regulatory axis that require further study. Finally, we evaluated the survival curves of the different subtypes for low and high miR-145 expression and observed a significant difference in both the classical and proneural subtypes but not in the mesenchymal subtype (Supplementary material, Fig. S6B). When we combined low miR-145 with high JAM-A, SOX2, and OCT4 in these subtypes, we observed that the median survival was consistently lower than their counterpart groups (Supplementary material, Table S2). Although only one group comparison reached significance (low miR-145/high SOX2 in the mesenchymal subtype), we noted that the lowest median survival was found in the proneural subtype, implying that these genes could be more informative for proneural GBM patient survival.

Fig. 6.
Low miR-145 levels are associated with poor patient prognosis, and proneural subtype has the lowest miR-145 levels. Survival curves in all glioblastoma patients in The Cancer Genome Atlas (TCGA) dataset show that low miR-145 levels are informative for ...


Adhesion is critical for many cellular functions including survival, proliferation, and migration. Adhesion is also essential for CSC maintenance because it promotes the abovementioned functions as well as localization to the niche and its instructive signaling. Therefore, the ability of a cell to acquire or lose adhesion mechanisms can result in transitioning into or out of the stem cell state; consequently, adhesion can be considered a hallmark of stemness. While adhesion molecules such as integrins and Id proteins have been successfully targeted in GBM,12,14,37 these same receptors are expressed in NPCs and prove challenging for clinical translation. JAM-A has recently been demonstrated to be essential for CSC maintenance and dispensable for NPC function,15 thereby representing a potential therapeutic target. The interactions presented in this manuscript corroborate the importance of JAM-A as its ectopic expression is capable of increasing stem cell frequency and proliferation. This effect, however, was differential in the specimens tested and could be due to differences in molecular subtype. Future studies would benefit from expanding this work to additional specimens of different genomic backgrounds to determine if this JAM-A/miR-145 regulatory axis is broadly applicable to CSC maintenance.

To uncover the mechanism by which self-renewal is coordinated by adhesion and integrated into a larger signaling network, we interrogated microRNAs and found evidence for a double feedback mechanism between JAM-A and miR-145, where the latter binds directly to the 3′UTR region of the JAM-A message, attenuating self-renewal. This miR-145 signaling system extends beyond JAM-A to core pluripotency factors (SOX2, OCT4, and NANOG), which were also downregulated upon miR-145 mimic introduction into CSCs. These data are consistent with previous reports that described miR-145 as a regulator of SOX2, OCT4, and KLF4 in human embryonic stem cells,32 miR-145 as a tumor suppressor in GBM via suppression of NEDD9,31 and a critical role for miR-145 in cell migration and self-renewal.38 We also described an association between miR-145 levels and patient survival based on data from TGCA. This showed that patients with lower levels of miR-145 (and in combination with high JAM-A levels) had a lower median survival than patients with high miR-145 levels. Despite the fact that this association was statistically significant, it should be tested in other independent cohorts and evaluated as a prognostic factor for patient survival. Similarly, future studies should analyze the ability of JAM-A to rescue the attenuated glioma-initiating capacity induced by ectopic expression of miR-145.

The implications of this signaling network extend beyond GBM as miR-145 regulates JAM-A expression in breast cancer,39 and more recently the expression of JAM-A was associated with decreased malignancy and invasiveness in malignant mesothelioma,40 bladder cancer,41 and lung adenocarcinoma.42 In addition to the link between JAM-A/miR-145 and pluripotency factors, JAM-A/miR-145 signals through the AKT pathway (a key CSC maintenance signaling node).34,35 The regulation of miR-145 by AKT may occur via activation of C/EBP-β as previously described.36 This interaction also likely explains why JAM-A overexpression is accompanied by activation of the AKT pathway and decreased miR-145 expression. However, miR-145 reduction may also be due to JAM-A activation via an adjacent cell, and this interaction represents a starting point for future inquiry on how JAM-A interaction between multiple cell types within the tumor (CSCs, non-CSCs, endothelial cells, immune cells) drives self-renewal. Taken together, our data demonstrate that CSCs possess specific mechanisms to preserve cell adhesion molecules and self-renewal genes that include a miR-145/JAM-A axis that drives CSC maintenance.


National Institutes of Health (NS083629 to J.D.L.). Work in the Lathia laboratory is also supported by National Institutes of Health grants CA198254, CA157948, and CA191263; a Distinguished Scientist Award from the Sontag Foundation; a Research Scholar Award from the American Cancer Society; Blast GBM; and Cleveland Clinic VeloSano Bike Race. Work in the Rich laboratory is supported by National Institutes of Health grants NS087913, CA154130, CA169117, CA171652, and NS089272 and the McDonnell Foundation. Work in the Vogelbaum laboratory is supported by the Wolf Family Foundation.

Supplementary Material

Supplementary Data:


We thank Grace Mullen and Jack Weisman for technical assistance and the members of the Lathia laboratory for constructive comments on the experimental design and manuscript. We thank Cathy Shemo, Patrick Barrett, and Sage O′Bryant for flow cytometry assistance.

Conflict of interest statement. None declared.


1. Reya T, Morrison SJ, Clarke MF, Weissman IL Stem cells, cancer, and cancer stem cells. Nature. 2001;414(6859):105–111. [PubMed]
2. Visvader JE, Lindeman GJ Cancer stem cells: current status and evolving complexities. Cell Stem Cell. 2012;10(6):717–728. [PubMed]
3. Galli R, Binda E, Orfanelli U et al. Isolation and characterization of tumorigenic, stem-like neural precursors from human glioblastoma. Cancer Res. 2004;64(19):7011–7021. [PubMed]
4. Hemmati HD, Nakano I, Lazareff JA et al. Cancerous stem cells can arise from pediatric brain tumors. Proc Natl Acad Sci USA. 2003;100(25):15178–15183. [PubMed]
5. Ignatova TN, Kukekov VG, Laywell ED, Suslov ON, Vrionis FD, Steindler DA Human cortical glial tumors contain neural stem-like cells expressing astroglial and neuronal markers in vitro. Glia. 2002;39(3):193–206. [PubMed]
6. Singh SK, Clarke ID, Terasaki M et al. Identification of a cancer stem cell in human brain tumors. Cancer Res. 2003;63(18):5821–5828. [PubMed]
7. Singh SK, Hawkins C, Clarke ID et al. Identification of human brain tumour initiating cells. Nature. 2004;432(7015):396–401. [PubMed]
8. Stupp R, Mason WP, van den Bent MJ et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. 2005;352(10):987–996. [PubMed]
9. Stupp R, Hegi ME, Mason WP et al. Effects of radiotherapy with concomitant and adjuvant temozolomide versus radiotherapy alone on survival in glioblastoma in a randomised phase III study: 5-year analysis of the EORTC-NCIC trial. Lancet Oncol. 2009;10(5):459–466. [PubMed]
10. Loulier K, Lathia JD, Marthiens V et al. beta1 integrin maintains integrity of the embryonic neocortical stem cell niche. PLoS Biol. 2009;7(8):e1000176. [PMC free article] [PubMed]
11. Niola F, Zhao X, Singh D et al. Id proteins synchronize stemness and anchorage to the niche of neural stem cells. Nat Cell Biol. 2012;14(5):477–487. [PMC free article] [PubMed]
12. Lathia JD, Gallagher J, Heddleston JM et al. Integrin alpha 6 regulates glioblastoma stem cells. Cell Stem Cell. 2010;6(5):421–432. [PMC free article] [PubMed]
13. Lathia JD, Li M, Hall PE et al. Laminin alpha 2 enables glioblastoma stem cell growth. Ann Neurol. 2012;72(5):766–778. [PMC free article] [PubMed]
14. Niola F, Zhao X, Singh D et al. Mesenchymal high-grade glioma is maintained by the ID-RAP1 axis. J Clin Invest. 2013;123(1):405–417. [PMC free article] [PubMed]
15. Lathia JD, Li M, Sinyuk M et al. High-throughput flow cytometry screening reveals a role for junctional adhesion molecule a as a cancer stem cell maintenance factor. Cell Rep. 2014;6(1):117–129. [PMC free article] [PubMed]
16. Kefas B, Godlewski J, Comeau L et al. microRNA-7 inhibits the epidermal growth factor receptor and the Akt pathway and is down-regulated in glioblastoma. Cancer Res. 2008;68(10):3566–3572. [PubMed]
17. Corsten MF, Miranda R, Kasmieh R, Krichevsky AM, Weissleder R, Shah K MicroRNA-21 knockdown disrupts glioma growth in vivo and displays synergistic cytotoxicity with neural precursor cell delivered S-TRAIL in human gliomas. Cancer Res. 2007;67(19):8994–9000. [PubMed]
18. Peruzzi P, Bronisz A, Nowicki MO et al. MicroRNA-128 coordinately targets Polycomb Repressor Complexes in glioma stem cells. Neuro Oncol. 2013;15(9):1212–1224. [PMC free article] [PubMed]
19. Lopez-Bertoni H, Lal B, Li A et al. DNMT-dependent suppression of microRNA regulates the induction of GBM tumor-propagating phenotype by Oct4 and Sox2. Oncogene. 2014;34(30):1994–4004. [PMC free article] [PubMed]
20. Siebzehnrubl FA, Silver DJ, Tugertimur B et al. The ZEB1 pathway links glioblastoma initiation, invasion and chemoresistance. EMBO Mol Med. 2013;5(8):1196–1212. [PMC free article] [PubMed]
21. Bao S, Wu Q, McLendon RE et al. Glioma stem cells promote radioresistance by preferential activation of the DNA damage response. Nature. 2006;444(7120):756–760. [PubMed]
22. Eyler CE, Wu Q, Yan K et al. Glioma stem cell proliferation and tumor growth are promoted by nitric oxide synthase-2. Cell. 2011;146(1):53–66. [PMC free article] [PubMed]
23. Li Z, Bao S, Wu Q et al. Hypoxia-inducible factors regulate tumorigenic capacity of glioma stem cells. Cancer Cell. 2009;15(6):501–513. [PMC free article] [PubMed]
24. Hu Y, Smyth GK ELDA: extreme limiting dilution analysis for comparing depleted and enriched populations in stem cell and other assays. J Immunol Methods. 2009;347(1–2):70–78. [PubMed]
25. John B, Enright AJ, Aravin A, Tuschl T, Sander C, Marks DS Human MicroRNA targets. PLoS Biol. 2004;2(11):e363. [PMC free article] [PubMed]
26. Marchenko S, Flanagan L Transfecting human neural stem cells with the Amaxa Nucleofector. J Vis Exp. 2007;6:240. [PubMed]
27. Cancer Genome Atlas Research Network. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature. 2008;455(7216):1061–1068. [PMC free article] [PubMed]
28. Dweep H, Sticht C, Pandey P, Gretz N miRWalk--database: prediction of possible miRNA binding sites by “walking” the genes of three genomes. J Biomed Inform. 2011; 44(5):839–847. [PubMed]
29. Zhang Y, Dutta A, Abounader R The role of microRNAs in glioma initiation and progression. Front Biosci. 2012;17:700–712. [PMC free article] [PubMed]
30. Moller HG, Rasmussen AP, Andersen HH, Johnsen KB, Henriksen M, Duroux M A systematic review of microRNA in glioblastoma multiforme: micro-modulators in the mesenchymal mode of migration and invasion. Mol Neurobiol. 2013;47(1):131–144. [PMC free article] [PubMed]
31. Speranza MC, Frattini V, Pisati F et al. NEDD9, a novel target of miR-145, increases the invasiveness of glioblastoma. Oncotarget. 2012;3(7):723–734. [PMC free article] [PubMed]
32. Xu N, Papagiannakopoulos T, Pan G, Thomson JA, Kosik KS MicroRNA-145 regulates OCT4, SOX2, and KLF4 and represses pluripotency in human embryonic stem cells. Cell. 2009;137(4):647–658. [PubMed]
33. Fang X, Yoon JG, Li L et al. The SOX2 response program in glioblastoma multiforme: an integrated ChIP-seq, expression microarray, and microRNA analysis. BMC Genomics. 2011;12:11. [PMC free article] [PubMed]
34. Bleau AM, Hambardzumyan D, Ozawa T et al. PTEN/PI3 K/Akt pathway regulates the side population phenotype and ABCG2 activity in glioma tumor stem-like cells. Cell Stem Cell. 2009;4(3):226–235. [PMC free article] [PubMed]
35. Eyler CE, Foo WC, LaFiura KM, McLendon RE, Hjelmeland AB, Rich JN Brain cancer stem cells display preferential sensitivity to Akt inhibition. Stem Cells. 2008;26(12):3027–3036. [PMC free article] [PubMed]
36. Sachdeva M, Liu Q, Cao J, Lu Z, Mo YY Negative regulation of miR-145 by C/EBP-beta through the Akt pathway in cancer cells. Nucleic Acids Res. 2012;40(14):6683–6692. [PMC free article] [PubMed]
37. Carbonell WS, DeLay M, Jahangiri A, Park CC, Aghi MK beta1 integrin targeting potentiates antiangiogenic therapy and inhibits the growth of bevacizumab-resistant glioblastoma. Cancer Res. 2013;73(10):3145–3154. [PMC free article] [PubMed]
38. Lee HK, Finniss S, Cazacu S et al. Mesenchymal stem cells deliver synthetic microRNA mimics to glioma cells and glioma stem cells and inhibit their cell migration and self-renewal. Oncotarget. 2013;4(2):346–361. [PMC free article] [PubMed]
39. Gotte M, Mohr C, Koo CY et al. miR-145-dependent targeting of junctional adhesion molecule A and modulation of fascin expression are associated with reduced breast cancer cell motility and invasiveness. Oncogene. 2010;29(50):6569–6580. [PubMed]
40. Cioce M, Ganci F, Canu V et al. Protumorigenic effects of mir-145 loss in malignant pleural mesothelioma. Oncogene. 2014;33(46):5319–5331. [PMC free article] [PubMed]
41. Kou B, Gao Y, Du C et al. miR-145 inhibits invasion of bladder cancer cells by targeting PAK1. Urol Oncol. 2014;32(6):846–854. [PubMed]
42. Hu J, Qiu M, Jiang F et al. MiR-145 regulates cancer stem-like properties and epithelial-to-mesenchymal transition in lung adenocarcinoma-initiating cells. Tumour Biol. 2014;35(9):8953–8961. [PubMed]

Articles from Neuro-Oncology are provided here courtesy of Society for Neuro-Oncology and Oxford University Press