Glioblastoma Stem Cells Respond to Differentiation Cues but Fail to Undergo Commitment and Terminal Cell-Cycle Arrest

Abstract

Glioblastoma (GBM) is an aggressive brain tumor whose growth is driven by stemcell-like cells. BMP signaling triggers cell-cycle exit and differentiation of GBM stemcells (GSCs) and, therefore, might have therapeutic value. However, the epigenetic mechanisms that accompany differentiation remain poorly defined. It is also unclear whether cell-cycle arrest is terminal. Herewe find only a subset ofGSCcultures exhibit astrocyte differentiation in response to BMP. Although overtly differentiated non-cycling astrocytes are generated, they remain vulnerable to cell-cycle re-entry and fail to appropriately reconfigure DNA methylation patterns. Chromatin accessibility mapping identified loci that failed to alter in response to BMP and these were enriched in SOX transcription factor-binding motifs. SOX transcription factors, therefore, may limit differentiation commitment. A similar propensity for cell-cycle re-entry and de-differentiation was observed in GSC-derived oligodendrocyte-like cells. These findings highlight significant obstacles to BMP-induced differentiation as therapy forGBM.

Graphical abstract

Figure 1

BMP Treatment Reduces Proliferation of GNS and NS Cells (A) Proliferation curves of seven GNS and two NS cells (NS-1 and NS-2). At day 7 all paired comparisons (GF versus BMP4) showed a significant difference in proliferation rate (p < 0.01). (B) Cells were expanded in the GFs EGF and FGF-2 (GF) or exposed to BMP4 in the absence of GFs for 8 days (BMP). Proliferation was assessed by EdU (16 hr incorporation) and astrocyte differentiation using GFAP (red). (C) Quantification of EdU-positive cells in proliferating conditions (GF), GF withdrawal (GF−), and BMP4 is shown. (D) Immunostaining for cell-cycle marker MCM2 (red) and quantification (bottom) are shown. (E) Relative mRNA expression levels of the BMPR1B in NS and GNS cell lines (fold change relative to normal brain). Error bars denote SD of two technical and two biological replicates for (C)–(E) (triplicates for immunostainings). Scale bars in (B) and (D), 100 μm.

Figure 2

Kinetics of DNA Methylation Changes Induced by BMP in NS and GNS Cells (A) The qRT-PCR analysis of G26 during a 48-day time course of BMP treatment. Error bars denote SD of four independent experiments (technical and biological duplicates). (B) A panel of 72 genes, previously defined as differentially expressed between GNS and NS cells (Engström et al., 2012), was analyzed using Taqman low-density arrays on G26 cells treated for up to 48 days with BMP4. (C) Total numbers of identified MVPs during BMP treatment for GNS cell lines (G19 and G26) and normal NS cells (NS-1) at each time point (BMP 8–48 days; B8, B16, B32, B48) are shown. (D) Dendrogram shows the 450K methylation array data.

Figure 3

Analysis of Sites of DNA Methylation Alterations Imposed by BMP Treatment (A) Percentage enrichment of epigenomic features (left), enhancers (middle), and PRC2 target genes (right) is shown (random resampling p value ≤ 0.001). (B) Visualization of MVPs over the time course of BMP is shown for the following: the astrocyte marker S100A6, the neurotransmitter receptor GRIK2, and the WNT-signaling tumor suppressor gene SFRP2. (C) GO analysis for the significantly altered MVPs with >30% change in methylation were analyzed using DAVID (GO BP_ALL). (D) Publicly available methylation datasets for pilocytic astrocytoma and human brain orbitofrontal cortex (non-neuronal) were used to generate a set of GBM-specific MVPs (top 100 MVPs shown), and these sites are shown for the BMP-treated GNS cells. Each experiment represents biological and technical duplicates of each sample.

Figure 4

RNA-Seq Analysis of BMP4-Treated G26 and NS-1 Cells (A) Gene expression of the MVP-associated genes GRIK2, SFRP2, and S100A6 (left) and the two top downregulated genes OLIG1 and OLIG2 (right). Fold change of the average of the number of reads in the two passages is shown. (B) Dendrogram of the RNA-seq data is shown. (C) The mRNA expression levels for many PRC2 target genes are frequently altered during BMP treatment. (D) The mRNA levels for FOXM1 and PLK1 are shown. (E) Heatmap shows transcription factors associated with the tumor-propagating state that recently was defined (Suvà et al., 2014). (F) Gene expression of DNA replication licensing proteins and cell-cycle regulators is shown relative to growth factors (GF) at day 0. (G) Quantification of MCM2-positive cells from immunocytochemistry. Each experiment represents biological and technical duplicates of each sample. Error bars denote SD of replicates.

Figure 5

BMP-Treated Cells Do Not Undergo Terminal Differentiation (A) G26 under self-renewal conditions (GF), treated with BMP4 for 7 (B7) and 54 days (B54), and when adding back GFs after the respective treatment times for 4 days (B7 + GF4 and B54 + GF4) is shown. (B) Quantitation of EdU-positive cells from (A). Error bars denote SD of five technical replicates. Student’s two-sided t test was used for pairwise comparisons (^∗p ≤ 0.05). (C) The qPCR analysis of G26 to determine the effects of re-exposure to GFs on astrocyte markers (AQP4 and GFAP) and NS cell-associated markers (EGFR, SOX2, and OLIG2). Error bars denote SD of two technical replicates from two biological replicates relative to cultures in GFs at day 0. (D) Dendrogram shows the 450K methylation data in G26 before and after the readdition of EGF and FGF-2 for 4 days (GF). (E) Kaplan-Meier curve showing survival data of mice transplanted with G26 grown under self-renewal conditions (GF), treated with BMP4 for 7 (B7) and 54 days (B54), and when adding back GFs after the respective treatment times for 4 days (B7 + GF4 and B54 + GF4) (n = 4–7 mice in each group). Scale bars, 100 μm.

Figure 6

ATAC-Seq Analysis of BMP4-Treated G26 and NS-1 Cells (A) PCA plot shows ATAC-seq data. (B) Heatmap generated for those loci that failed to be closed during BMP-induced differentiation in G26 compared to normal NS cells. KEGG glioma gene proximal loci (<10 kb) are labeled in red; cell-cycle genes in blue are the shared glioma cell-cycle loci in black. (C) Genome browser view of the IGF1 glioma locus is shown. (D) Motif enrichment analysis identified a high frequency of SOX motifs within those peaks identified in (B); two top motifs (lowest adjusted p values) for SOX2 are shown (see also Table S1 for full motif sets).

Figure 7

Oligodendrocyte-like Cells Generated from GNS Cells Fail to Terminally Differentiate (A) G144 cells stained with O4 (red) and EdU under self-renewal conditions (+GF) and under differentiating conditions (GF withdrawal, −GF) (left) and flow cytometry (right) are shown. (B) Live O4 immunostaining in G19, G25, and G144 after GF withdrawal for 7 days is shown (green). (C) Quantitation of flow analysis from (B). Error bars denote SD of biological replicates (n = 3–7). (D) The qPCR analysis of the oligodendrocyte markers MBP and MAG. Error bars denote SD of biological and technical duplicates. (E) EdU incorporation in proliferating G26 (left), after 10 days GF withdrawal (middle), and at re-exposure to EGF and FGF-2 (GF) (right). Quantification shows the same experiment and additional experiments at 20 days (six biological replicates). (F) (Top) Experimental design for clonal analysis of differentiation commitment in O4-positive cells. (Bottom) Live immunostaining of O4 in the G144 culture after GF withdrawal (left), at single-cell deposition (middle), and after adding back GFs (right) is shown. (G) Quantification of O4-positive cells after single-cell deposition indicates that O4-positive cells can re-enter cell cycle and proliferate. Total number of cells in each well was scored after 10 days of GF re-addition. Undifferentiated controls, UD; cell death, cross. Scale bars, 100 μm.