Fig 2: Target dependent effects of miR-449a on proliferation, differentiation and glioma biology (a) in stemness—maintaining growth conditions with EGF, FGF enriched serum-free medium, miR-449a—mediated effects are predominantly exerted through CCND1 inhibition, thus reducing invasion and proliferation. For example, miR-449ahigh Rb/p53 cells have low expression levels of Ccdn1, proliferate and migrate slower, and have less self-renewal capacity. b in growth conditions promoting neural phenotypes in vitro, such as FBS, Forskolin, retinoic acid and in vivo experimental settings and humans gliomas, miR-449a suppresses GPR158, reducing neural marker expression, and is associated with higher glioma grade and shorter survival. Numbers and letters provide a reference to figures in the text

Fig 3: Patients with gliomas expressing higher level of GPR158 survive longer and show better response to chemotherapy. a Kaplan–Meier estimates of overall survival in all grades of gliomas from our institution, grouped by miR-449a expression levels or by (b) GPR158 expression measured by staining intensity. Patients with tumours expressing low levels of miR-449a or high levels of GPR158 survive significantly longer. The 'miR-449a low' and 'miR-449a high' groups, and the 'GPR158 low' and 'GPR158 high' groups were separated by the median of expression levels or staining intensity. c Stratification of survival according to GPR158 expression levels (rsem) in the TCGA cohort. The survival pattern from TCGA data confirms the finding in (a). d The cohort of patients (from TCGA) with IDH mutant tumours expressing high levels of GPR158 show a trend of longer survival than low expressors. However, in this cohort the difference does not reach significance level. e Proneural and f neural subtypes of GBMs (from TCGA) with higher expression of GPR158 show a significantly longer survival, whilst no difference is seen in GBM of (g) classical (h) or mesenchymal profiles. i GBM with high GPR158 expression show a significantly better response to chemotherapy. All p values were estimated using log-rank test. j Low grade or k high grade gliomas, stratified by CCND1 expression do not show a difference in survival, indicating a role mediated by the miR-449a target GPR.158 rather than CCND1

Fig 4: Identification of Gpr158 as a direct target of miR-449a. a Venn diagram with eight candidate genes emanating from 101 in silico putative targets and 1000 down-regulated genes in experimental PNETs compared with gliomas by analysis of exon expression array. b Candidate gene expression level is validated by RT-qPCR in Rb/p53 (orange error bars), Rb/p53ant (red bars) and Pten/p53 cells (grey bars). Most differentially expressed Ccnd1 and Gpr158 are further analysed, as their expression is similar in Rb/p53ant and Pten/p53 cells, but significantly higher than in Rb/p53 cells. c IHC staining shows that Gpr158 expression is minimal in miR-449 highly expressing PNETs, but strong in miR-449 low expressing gliomas. Scale bar 50 µm. d Schematic illustration of Ago2 and biotin double pull-down assay for assessment of miRNA-mRNA binding. Commercial synthetic miR-449a mimics are transfected into neural stem cells, and Ago2 immunoprecipitation is carried out to confirm that miRNA-mRNA binding is RISC dependent. Fraction 1 represents the input RNA, fraction 2 the Ago2 depleted fraction, i.e, miRNA and mRNA unbound to Ago2. Fraction 3 represents miRNA449a-mRNA complex bound to Ago2, representing the degradation complex RISC. These fractions were then tested for the enrichment of Gpr158 and Ccnd1 transcripts: e Enrichment of Ccnd1 and Gpr158 is measured after pull-down using RT-qPCR. The x axis shows the fraction as described in (d). There is a highly significant enrichment in fraction 3 (Ago2-dependent miR-449a –Gpr158 complex) indicating direct interaction. f miR-449a binding sequence in the 3′ UTR of Gpr158. A mutation of the 3’UTR of Gpr158 generated in the site complementary to the seed region of miR-449a. *Indicates the mutant nucleotides. g miR-449a directly targets Gpr158 by interacting with its 3′ UTR. Relative luciferase activity (normalized to control) of BTSCs transfected with pMIR-Gpr158-3′ UTR-wt or pMIR-Gpr158-3′ UTR-mut, and co-transfected with miRNA negative control or miR-449a mimics. This suggests a significant miR-449a mediated downregulation of Gpr158, which is not seen in the mutant control. All figures: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001 (Student’s t-test). Each bar represents mean ± sd

Fig 5: miR-449a antagonises Gpr158-induced neural differentiation. a Induction of neural differentiation with 3% FBS increases Gpr158 and reduces miR-449a expression. Ccnd1 expression in EGF, FGF-enriched stem cell medium (upper part) is suppressed in miR-449ahigh Rb/p53 with fewer nuclei labelled, and the overall expression is strongly reduces upon growth in 3% FBS supplemented medium. Quantification of miR-449a and Gpr158 expression in these cells on the right. b Stable overexpression of Gpr158 antagonises miR-449a and promotes neural differentiation in miR-449ahigh Rb/p53 mBTSC. Upper panel, cells transformed with lentivirus expressing GFP or an Gpr158 expressing lentivirus, stained for doublecortin to label postmitotic neural progenitor cells and early immature neurons in serum-free, EGF and FGF enriched stem cell medium. RT-PCR profiling shows expression of 'proneural' and downregulation of 'mesenchymal' genes upon Gpr158 expression (c) Knock-down of miR-449a or overexpression Gpr158 induce glial and neural marker expression. Rb/p53 mBTSC (miR-449ahigh), were either transfected with vehicle (pcDNA 3.1+, c1), miR-449a antagomir (GFP labelled, c3) or the lentivirus pLX301-Gpr158 (c5). Following exposure to 3% FBS for 48 h (right panel, c2, 4, 6), only miR-449alow cells (c4) or increase of GPR158 levels (b6) but not vector-only transfected controls (b2) show increased GFAP and DCX positive mBTSC. d The opposite result is seen with Pten/p53 mBTSC (miR-449alow) are treated with miR-449a mimic or Gpr158 siRNA. In EGF, FGF enriched, serum free medium only rare DCX positive mBTSC are seen (d1). Exposure to 3%FBS enriches in GFAP or DCX expressing mBTSC (d2). Transfection with GFP labelled miR-449a (d3, 4) mimic greatly reduces the number of GFAP and DCX positive cells in 3% FBS enriched medium (d4), compared to d2. The DCX labelled differentiated mBTSC (d4 inset) was spared from (GFP labelled) miR-449a mimic. In keeping, knockdown of Gpr158 abolishes GFAP and DCX expression on mBTSC (c6). e quantification of DCX (red bars) and GFAP (green bars) positive cells in miR-449a antagomir or GPR158 overexpressing Pten/p53 mBTSC. E, quantification of DCX (red bars) and GFAP (green bars) positive cells in miR-449a mimic or GPR158 knock-down Rb/p53 mBTSC