Fig 2: HSP90 enhances SMAD2, SMAD3, and TβRII expression and TGF-β1 pathway activation in Müller glial cells. (A) MIO-M1 Müller glia cells were transduced with lentiviruses encoding for the human FLAG-tagged HSP90 or GFP (as control). Transduction efficiency was monitored by fluorescence microscopy analysis of GFP expression in control cells. (B) Transduced cells were treated with human recombinant TGF-β1. Expression of endogenous SMAD2, SMAD3, and TGFβRII and transduced FLAG-tagged HSP90 were analyzed by immunoblot. TGF-β1 pathway activation was assayed using a specific phospho-SMAD2 (Ser465/467) antibody. The expression of β-actin was used as loading protein control.

Fig 3: Smad3 constitutes an important effector in both proneural and mesenchymal GSCs.A Heatmap showing the molecular subtype marker expression in proneural GSCs (PN12, PN16, and PN19), and mesenchymal GSCs (ME23, ME27, and ME29) with or without TGF-β1 treatment for 4 or 8 h. Proneural markers: DLL3, OLIG2, ASCL1, CD133, and SOX2. Mesenchymal markers: CD44, CHI3L1, TIMP1, BCL3, COL3A1, MGP, and POSTN. Z-scores were calculated from the △Ct values obtained in the quantitative polymerase chain reaction analysis. B Western blotting analysis of SMAD2, p-SMAD2, SMAD3, and p-SMAD3 expression in proneural GSCs (PN12, PN16, and PN19) and mesenchymal GSCs (ME23, ME27, and ME29) with or without TGF-β1 treatment for 8 h. C Kaplan–Meier survival curves for patients with mesenchymal GBM and a high Smad3 gene signature (n = 44) and those with a low Smad3 gene signature (n = 33). D Protein–protein interaction networks indicated by STRING and Cytoscape software show overlapping peaks in both proneural and mesenchymal GSCs and also identify those that are unique to a particular subtype of GSCs or shared between the proneural and mesenchymal GSCs. E Lysates from proneural GSCs and mesenchymal GSCs were subjected to immunoprecipitation using anti-Smad3 antibody and then immunoblotted with anti-E2F4 and anti-EP300 antibodies. GSCs glioblastoma stem cells.

Fig 4: Inhibition of HIF-1α or Smad3 could reduce cell viability and self-renewal in the proneural and mesenchymal subtypes of GSCs.A Heatmap showing molecular subtype marker expression in proneural GSCs (PN12, PN16, and PN19), mesenchymal GSCs (ME23, ME27, and ME29), and proneural (or mesenchymal) GSCs infected with the indicated plasmids or shRNA or treated with the indicated inhibitors. HIF-1α inhibitor, GN44028, 12.5 μM. Palmitoylation inhibitor, 2-BP, 100 μM. Smad3 inhibitor, (E)-SIS3, 20 μM. EP300 inhibitor, C646, 15 μM. Z-scores were calculated from the △Ct values obtained in the quantitative polymerase chain reaction analysis. B Viability of proneural PN12 GSCs transfected with the indicated plasmids or shRNA or treated with GN44028, C646, 2-BP, or (E)-SIS3 for 48 h. C A serial dilution assay was used to determine the self-renewal capacity of PN12 GSCs transfected with the indicated plasmids or shRNA or treated with GN44028, C646, 2-BP, or (E)-SIS3. D Viability of mesenchymal ME23 GSCs transfected with the indicated plasmids or shRNA or treated with GN44028, C646, 2-BP, or (E)-SIS3 for 48 h. E A serial dilution assay was used to determine the self-renewal capacity of ME23 GSCs transfected with the indicated plasmids or shRNA or treated with GN44028, C646, 2-BP, or (E)-SIS3. GSCs glioblastoma stem cells, HIF-1α hypoxia-inducible factor 1-alpha.

Fig 5: TGFB2-AS1 Interacts with EED(A) Predicted secondary structure of TGFB2-AS1 generated by RNAfold. The 5' and 3' ends and break-points (arrowheads) of fragments used in RNA pull-down are marked with nucleotide numbers: 1–200, 201–395, and 396–557. Each nucleotide is color-coded (see color scale): dark blue (probability 0) indicates the lowest and dark red (probability 1) the highest base-pairing probability; for unpaired regions, the color highlights the probability of being unpaired.(B) Schematic illustration of the RNA pull-down assay followed by mass spectrometry analysis.(C) In vitro transcribed biotinylated F-luciferase (F-luc) mRNA and TGFB2-AS1 RNA analyzed by agarose gel electrophoresis and molecular size markers (nt).(D) Total protein numbers and examples of proteins interacting with TGFB2-AS1 or F-luc mRNA, identified by mass spectrometry in nuclear lysates of HaCaT cells, stimulated with or without TGF-ß for 1 h. EED is highlighted (red) and Smad3 is in brackets because of low statistical coverage.(E) RNA pull-down assay using biotinylated full-length (FL) or different TGFB2-AS1 fragments immobilized on streptavidin beads and lysates from HA-EED overexpressing HEK293T cells. The same cell lysate was applied to each specific RNA. Biotinylated TGFB2-AS1 RNA fragments were analyzed by agarose gel electrophoresis (top). The proteins retained on the RNA beads and total cell lysates (TCLs) were analyzed by immunoblotting using the indicated antibody. Representative immunoblots out of two independent experiments with molecular size markers (kDa).(F) In vitro transcribed biotinylated anti-TGFB2-AS1 and TGFB2-AS1 RNAs were analyzed using agarose gel electrophoresis along with molecular size markers (nt).(G) Quantitative real-time PCR to determine mRNA levels of EED in HEK293T cells transiently transfected with siEED or siControl. The error bars represent SD from three different experiments.(H) RNA pull-down assay using biotinylated (biot-) anti-TGFB2-AS1 and TGFB2-AS1 RNAs (black background) immobilized on streptavidin beads and lysates from HEK293T cells transiently transfected with control (-) or specific siRNA targeting EED (siEED; +) and overexpressing HA-EED, HA-EZH2, and HA-SUZ12. The same cell lysate was applied to each specific RNA. The proteins retained on the RNA beads and total cell lysates (TCLs) were analyzed by immunoblotting using the indicated antibodies. Representative immunoblots out of three independent experiments, with two different exposures (exp.) for HA-EED and molecular size markers (kDa). A star indicates a non-specific protein band. Dotted lines serve orientation and alignment (the immunoblots were not disrupted).(I) RIP of HaCaT lysates using antibodies against endogenous EED, EZH2, SUZ12, or control IgG. Error bars represent SD from three different experiments. Corresponding immunoblots indicate the immunoprecipitated IgG or specific protein (marked by arrowheads) and molecular size markers (kDa).(J) RIP of HaCaT lysates after transient transfection with siEED or siControl, using antibody against endogenous EZH2 or control IgG. Error bars represent SD from three different experiments.(K) Immunoprecipitation of endogenous SUZ12 (or control IgG) in HEK293T cells transiently transfected with the indicated expression constructs and stimulated or not with TGF-ß1 for 1 h, followed by immunoblotting for endogenous SUZ12, endogenous EZH2, or transfected HA-EED. Immunoblots of corresponding total cell lysates (TCLs) for the same three proteins in addition to p-Smad2 (indicator of TGF-ß stimulation), total Smad2/3, and loading control ß-actin, along with molecular size markers (kDa). Stars indicate non-specific protein bands. Representative immunoblots out of three independent experiments are shown.