Identification of miRNAs in high-grade gliomas
To identify miRNAs that are recurrently deregulated in high-grade gliomas, we used quantitative RT-PCR to profile expression of 192 miRNAs in human non-neoplastic brain tissues, AAs and GBMs. From our analyses, we identified a number of miRNAs that have been described previously in GBM tumors such as miR-10b (see ) and the apoptosis regulator miR-21 (see [21, 22, 31]). We also identified a number of miRNAs, including miR-124 and miR-137, which have not been described in prior GBM profiling studies. It remains unclear why miR-124 and miR-137 were not detected previously in GBM tumors, particularly in light of our results that show dramatic expression decreases of miR-124 and miR-137 in GBMs (and AAs) relative to non-neoplastic brain tissue, and results that show clear down-regulation of miR-124 expression in human oligodendrogliomas , human astroblastomas  and GBM cell lines [32, 33].
Another notable discrepancy is that of miR-221 expression, which was not overexpressed in any of the tumors tested in our study, but was shown to be overexpressed in five out of nine of GBMs studied by Ciafre et al. , and has been shown to inhibit expression of the cell-cycle inhibitor p27(Kip1) in GBM cells . The most obvious differences between our study and previous GBM profiling studies are: (1) the control tissues, adult glioses from epileptic surgeries, a routinely used control tissue for GBM molecular profiling studies, versus normal human fetal brain and macroscopically characterized surgical specimens from the tumor periphery [21, 22]; (2) profiling technologies, TaqMan versus micro-array [21, 22]. Future profiling studies on larger numbers of patient samples will help to resolve these apparent discrepancies.
Our expression analyses revealed a total of 35 miRNAs that were differentially expressed between high-grade gliomas and non-neoplastic brain tissue (P < 0.05, Table 1). A vast majority of these miRNAs were down-regulated (29 out of 35; 83%), which is consistent with observations that miRNA expression is globally down-regulated in multiple tumor types . Of the 35 miRNAs, we identified six HGA-miRNAs, which were down-regulated in both AA and GBM tumors at a more stringent degree of significance (P < 0.01): miR-7, miR-124, miR-129, miR-137, miR-139 and miR-218. Although we restricted further analyses of these six miRNAs to miR-124 and miR-137 because of their elevated expression during adult NSC differentiation (Figure 1B), assessments of the other HGA-miRNAs may lead to novel insights into the biology of high-grade gliomas. Similarly, assessments of the miRNAs that were differentially expressed in AA tumors only or GBM tumors only (Table 1) may shed light on the biological differences underlying these different tumor grades.
It is important to note that our miRNA expression profiling studies were conducted at the tissue level, not at the cellular level, which has important implications for the interpretation of our results. In particular, prior work  has shown that miR-124 is only expressed in the neurons of adult human brains, which indicates that our observed decrease in miR-124 expression in HGAs is a likely consequence of there being relatively fewer neurons in tumor tissue compared with non-neoplastic glioses controls. While this does not change our conclusions that miR-124 and miR-137 can induce mNSC-, mOSC- and human GBM-derived stem cell (hGSC)-differentiation, it indicates that in situ expression analyses of miRNAs in HGAs, non-neoplastic adult brain tissue, and during fetal- and post-natal development of the mammalian central nervous system will be an important component of studies aimed at investigating the functions of miRNAs during normal brain development and tumorigenesis.
We also note that in this study we analyzed 192 of the 533 known human miRNAs that are currently described in miRBase, release 10.0 (see ), which reflects the rapid pace of miRNA discovery since the inception of our miRNA expression studies. It is likely that miRNAs of potential significance to brain tumor biology have not been assessed here. Examples of such miRNAs include those that show enriched expression in brain tissue such as miR-451 and miR-488 (see ) and miRNAs that have been implicated in the etiology of other tumor types, such as miR-346 in follicular thyroid carcinoma . Therefore, comprehensive miRNA expression studies are warranted in large HGA tumor sets that are linked to clinical data, such as survival and therapeutic response in order to generate an in-depth assessment of the role of miRNAs in brain cancer etiology and therapy.
Regulation of miR-124 and miR-137 expression
Our results reveal two potential mechanisms by which miR-124 and miR-137 may be suppressed in stem cells and/or tumor cells. The first mechanism is growth factor signaling: removal of EGF, and FGF from the culture media resulted in robust increases in miR-124 and miR-137 expression in adult NSCs. Given that activation of EGF , PDGF  and FGF  signaling pathways have each been implicated in gliomagenesis, it is reasonable to speculate that one mechanism by which growth factor signaling promotes brain tumor formation is through suppression of miR-124 and/or miR-137 expression and NSC/TSC differentiation. Further analyses are required to determine the relative contributions of EGF-, FGF- and PDGF-induced signaling on suppression of miR-124 and miR-137 transcription in adult NSCs and GBM tumor stem cells.
The second mechanism by which miR-124 and miR-137 expression may be suppressed in GBM stem cells is via epigenetic modification of their transcriptional regulatory sequences. Indeed, epigenetic modification of specific miRNAs in other tumor types has been reported recently. For example, miR-127, which is down-regulated in prostate, colon and bladder tumors relative to matched normal tissues, is up-regulated in cell lines derived from these tumor types following inhibition of DNA demethylation and histone deacetylase . Of particular interest to our studies, miR-124 is hyper-methylated in over one-third of colon, breast, lung, lymphoma and leukemia primary tumors, and is up-regulated in breast (MCF-7) and colon (HCT-116) cancer cell lines following DNA demethylation . We observed that miR-137 expression increased in GBM cell lines U87 and U251 following treatment with the DNA demethylating agent 5-aza-dC (Figure 1B). Interestingly, we did not observe an increase in miR-124 expression in either cell line following 5-aza-dC treatment. Further analyses of miR-137 and miR-124 promoter sequence methylation in primary tumors, TSCs and NSCs are warranted to establish the degree to which epigenetic mechanisms contribute to suppression of these miRNAs in HGAs.
Regulation of differentiation and the cell cycle by miR-124 and miR-137
Previous studies have demonstrated that miR-124 is up-regulated during development of the rodent nervous system [41, 42], and during neuronal differentiation of mouse ES cells , and mouse and human embryonal carcinoma cells . Further, neuronal differentiation is enhanced following ectopic overexpression of miR-124 in mouse ES cells , mouse neuroblastoma cells , and mouse embryonal carcinoma cells . Our results indicate that overexpression of either miR-124 or miR-137 promotes neuron-like differentiation of non-neoplastic adult (mNSCs), mOSCs and CD133+ hGSCs. Thus, our study is the first to implicate miR-124 in neuronal differentiation of post-natal NCSs and brain TSCs.
The ability of miR-124 to induce robust stem cell differentiation appears to be dependent on cell type, developmental timing and other, as yet unidentified, factors. For example, in mouse neuroblastoma cell lines CAD and Neuro2a, ectopic up-regulation of miR-124 alone is sufficient to induce neuron-like differentiation, whereas in mouse embryonic carcinoma cells (P19), miR-124 enhances neuronal differentiation only in the presence of retinoic acid, an established inducer of P19 neuronal differentiation . Investigations of miR-124 expression and function during development of the embryonic chick spinal cord have determined that the proneural activity of miR-124 is, at best, subtle [43, 44], suggesting that additional factors- and/or signals are required for robust neurogenesis at this developmental stage. Our studies show that miR-124 and miR-137 enhance neurogenesis of mNSCs, mOSCs and hGSCs in the absence of growth factor signaling. Although we have not tested whether miR-124 and miR-137 alone can induce differentiation of the various stem cells tested in this study, transfection of miR-124 or miR-137 alone was sufficient to induce G1 cell cycle arrest in standard GBM cell lines (Figure 5A). However, cell cycle arrest was more pronounced in miR-124- and miR-137-transfected GBM cells (SF6969) that were deprived of growth factors (Figure 5B). Overall, the most robust effects of miR-124 and miR-137 overexpression on cellular differentiation and proliferation were observed in growth factor-deprived human cells (Figures 4B and 5B). Collectively, our results suggest that while miR-124 and miR-137 have the capacity to induce alone cell cycle arrest and differentiation in human GBM cells and stem cells, abrogation of growth factor signaling enhances their capacity to do so. Additional studies will be required to address this hypothesis, and incorporation of additional GBM and oligodendroglioma-neurosphere lines will be required to address the general applicability of our results in relation to the biology and therapeutics of these diseases.
Recent studies have begun to shed light on the molecular mechanisms by which miR-124 regulates differentiation and proliferation. For example, miR-124 directly targets PTBP1 (PTB/hnRNP I) mRNA, a global repressor of alternative pre-mRNA splicing in non-neuronal cells, resulting in the transition from non-neuronal- to neuronal-specific alternative splicing patterns . miR-124 also directly targets and suppresses expression of small C-terminal domain phosphatase 1 (SCP1), an inhibitor of neuronal gene expression . Finally, miR-124 overexpression in HCT-116 colon cancer cells inhibits the expression of CDK6, an established target of miR-124 (see ). Our studies revealed that miR-137, as well as miR-124, inhibited expression of CDK6, a predicted target of both miRNAs. Further, as with miR-124a (see ), our results show that miR-137 is a direct inhibitor of CDK6. Overexpression of miR-124 or miR-137 also reduced the expression of phosphorylated RB (Figure 6B), a downstream target of CDK6 . It is interesting to note that CDK6 is known to regulate both cell cycle progression and differentiation (reviewed in ), suggesting that mir-124- and miR-137-mediated inhibition of CDK6 may, in part, account for the observed effects on GBM cell proliferation and differentiation in this study. Further investigations are needed to define the relationship between CDK6 down-regulation and cell cycle arrest and/or differentiation in GBM stem cells, and to identify and characterize additional miR-124 and miR-137 target genes.
Therapeutic potential of miR-124 and miR-137
The ability of miR-124 and miR-137 to induce potent antiproliferative and prodifferentiation effects in CD133+ and CD133- human GBM cells suggests their potential value for treatment of this disease. RNAi-based therapeutics holds great promise for the development of entirely novel therapeutic strategies for disease treatment , and early phase clinical trials using siRNAs are currently underway . While delivery of siRNAs or miRNAs to the central nervous system is particularly challenging because of the blood brain barrier, a number of promising strategies have been developed recently to circumvent this problem. These include intranasal delivery of oligonucleotides , lipid encapsulation and targeted delivery of nucleic acids [48, 49], and direct administration of therapeutic agents to brain tumor tissues by convection-enhanced delivery [50, 51]. Further testing of miR-124 and miR-137 in pre-clinical models of GBM [52, 53] in conjunction with various delivery strategies will help define their ultimate therapeutic potential for treatment of GBM.