Nevertheless, triple immunofluorescence staining revealed that our differentiation method induced the expression of both astrocytic and neuronal markers in the differentiating GICs simultaneously. These results 
might suggest a differentiation of these cells towards the neuronal lineage, but retaining the expression of GFAP that is usually restricted to neural precursors in the neuronal lineage, while it is abundantly expressed in the astrocytic lineage. This aberrant marker expression in differentiating GICs has been previously reported by other groups. Analogously to what has been reported for the differentiation of normal neurons, most of the miRNAs that changed their expression levels upon GIC differentiation in our model belong to the same miRNA clusters, and several Rapamycin paralog clusters are involved. For instance, the paralog miRNA clusters miR-106a/363, miR-106b/25 and miR-17/92 are down-regulated upon differentiation, while clusters miR-29a/29b and miR221/222 are strongly upregulated, suggesting an important role for coordinate regulatory miRNA networks during GIC differentiation. To assess the significance of these two up-regulated miRNA clusters in the differentiation process, we performed transfection experiments using precursors or inhibitors of these miRNAs and analyzed the expression of differentiation markers. Cluster miR-29a/29b did not induce the expression of the studied differentiation markers, but sensitized the cells to apoptosis by targeting MCL1, a bona-fide target of the miR-29 family. Interestingly, MCL1 is the most over-expressed protein of the BCL2 family in the majority of malignant gliomas, and neutralization of MCL1 in glioma cells has been reported to induce apoptosis and increase chemotherapy-induced apoptosis, suggesting that miR-29a/29b over-expression could be studied as a possible therapy for GBM. The upregulation of cluster miR-221/222 that we observed upon GIC differentiation is more controversial, since this cluster has been found over-expressed in GBM compared to non-transformed tissue, being particularly associated to the astrocytic GBM subclass. Conversely, both miRNAs have been shown to inhibit proliferation in the TF-1 erythroleukemic cell line and to reduce the stem cell repopulating activity of cord blood CD34+ cells through inhibition of KIT. Of note, KIT amplification is a frequent Adriamycin Topoisomerase inhibitor alteration in GBM. Thus, these miRNAs probably can exert pro-oncogenic or tumor suppressor functions depending on the cellular context. Regarding neural cell differentiation, miR-221 has been found highly up-regulated upon nerve growth factor-induced neuronal-like differentiation of PC12 rat pheochromocytoma cells. miR-221 could be exerting a similar role during GIC differentiation. One of the most surprising findings of this work is the prodifferentiation role of miR-21 over-expression in GICs. miR-21 is regarded as an onco-miR in GBM, as well as in other tumors, and its over-expression has been associated to poor clinical outcome. Indeed, miR-21 has shown a widespread involvement in the inhibition of tumor suppressor genes in GBM cells, targeting multiple components of the p53, transforming growth factor-�� and mitochondrial apoptosis pathways. Consequently, the inhibition of miR-21 expression with therapeutic intent has been suggested as a possible treatment for GBM and some preliminary in vitro and in vivo assays have provided promising results. For instance, it has been reported that the inhibition of miR-21 in GBM cells as well as in glioma xenotransplant-bearing mice promotes apoptotic cell death of the tumor cells, but no studies of its effects on survival were performed, as all animals were sacrificed 6 days after treatment with LNA-anti-miR-21.