The ineffectiveness of standard therapies in the treatment of malignant gliomas signals an urgent need for the development of disease-relevant targeted agents. This process itself requires a more complete understanding of underlying glioma biology, particularly with regard to its well-established variability at the molecular level. Several miRNA-mediated networks have been directly implicated in gliomagenesis, and recent work has shown that robust miRNA signatures designate distinct GBM subclasses, highlighting the relevance of miRNA biology to the analysis of molecular heterogeneity in malignant glioma 
. In the present study, we aimed to identify and characterize miRNAs involved in the pathogenesis of proneural gliomas using an approach that repeatedly combined in vitro
methodologies with existing bioinformatic resources. As such, the ready availability of TCGA profiling data for GBM was an essential component of our basic strategy. Our findings indicate that proneural gliomas are specifically characterized by miR-34a downregulation with subsequent derepression of the miRNA's downstream target PDGFRA, a process that promotes tumorigenesis both in vitro
and in vivo
. The restriction of this regulatory interaction to proneural gliomas is perhaps not surprising given the highly distinctive molecular profiles exhibited by different GBM expression subclasses, and the central role of PDGFRA in proneural gliomagenesis. miRNA behavior and target profiles are known to be highly dependent on cellular context. Indeed, specific miRNAs have even been shown to behave as both oncogenes and tumor suppressors in different cancer variants 
miR-34a was originally identified as a likely tumor suppressor miRNA and a downstream transcriptional target of p53 
. Prior reports have shown that miR-34a is downregulated in GBM compared to normal brain, and that it inhibits cell proliferation, survival, and invasion in adherent glioma cell lines by targeting MET, NOTCH1, NOTCH2, and CDK6 
. Additionally, miR-34a appears to promote differentiation in glioma cells grown in stem-like conditions 
. Our findings indicate that PDGFRA represents a crucial target of miR-34a in the setting of proneural gliomagenesis. Our examination of other transcripts known to interact with miR-34a in GBM revealed that only NOTCH1 was significantly repressed by the miRNA in bona fide
proneural glioma cells. Moreover, specific siRNA knockdown of NOTCH1 failed to demonstrate functional consequences, while PDGFRA knockdown yielded significantly reduced BrdU incorporation, emphasizing functional relevance. Nevertheless, we anticipate that other important miR-34a targets remain to be identified, particularly given that selective PDGFRA knockdown does not fully recapitulate the effects of miR-34a on cell proliferation in proneural glioma cells.
miR-34a repression in proneural gliomas appears to result directly from enhanced PDGF signaling. We found that constitutive activity in the PDGF pathway directly downregulates miR-34a expression in a manner that is completely reversible by imatinib administration. The well-established role of p53 in the transcriptional activation of miR-34a prompted us to investigate whether miR-34a repression in proneural gliomas is mediated through a p53-dependent mechanism. Indeed, earlier work has identified a potential conduit for such transcriptional regulation through p-AKT and p-MDM2 
. However, our western blot analysis in proneural TS543 cells failed to demonstrate significant changes in p53 levels in response to imatinib, despite robust modulation of both p-AKT and p-MDM2 levels. These finding indicate, somewhat surprisingly, that upstream regulation of miR-34a is, at best, only partially regulated by a p53-dependent mechanism and that alternative molecular pathways are likely involved in proneural GBMs. Consistent with this conjecture are TCGA data demonstrating that proneural subclass, rather than p53 genomic status, is most predictive of miR-34a expression levels. Additionally, the fact that miR-34b and -34c, which are also regulated by p53 
, are not similarly repressed by dysregulated PDGF signaling provides further support for the contribution of p53-independent mechanisms.
In summary, we identify a miRNA-mediated network that promotes tumorigenesis in a specific glioma subtype by deregulating that subtype's single most defining oncogenic driver. Disrupting the homeostatic equilibrium between miR-34a and PDGFRA potentially gives rise to a cancer-promoting, feed-forward loop, one that could ultimately lead to both cell proliferation and tumor formation. Intriguingly, such a mechanism could conceivably drive gliomagenesis even in the absence of canonical genomic aberrations (e.g. mutations or copy number gains) in core signaling pathway components. This consideration is particularly relevant to proneural gliomas, which, despite their strong associations with dysregulated PDGF signaling, only harbor mutations and or amplifications of PDGFRA in a minority of cases. On a related note, these findings also serve to further emphasize the therapeutic potential of inhibiting PDGF signaling in the proneural subtype of malignant glioma.