Epigenetic Activation of WNT5A Drives Glioblastoma Stem Cell Differentiation and Invasive Growth

Abstract

Glioblastoma stem cells (GSCs) are implicated in tumor neovascularization, invasiveness, and therapeutic resistance. To illuminate mechanisms governing these hallmark features, we developed a de novo glioblastoma multiforme (GBM) model derived from immortalized human neural stem/progenitor cells (hNSCs) to enable precise system-level comparisons of pre-malignant and oncogene-induced malignant states of NSCs. Integrated transcriptomic and epigenomic analyses uncovered a PAX6/DLX5 transcriptional program driving WNT5A-mediated GSC differentiation into endothelial-like cells (GdECs). GdECs recruit existing endothelial cells to promote peritumoral satellite lesions, which serve as a niche supporting the growth of invasive glioma cells away from the primary tumor. Clinical data reveal higher WNT5A and GdECs expression in peritumoral and recurrent GBMs relative to matched intratumoral and primary GBMs, respectively, supporting WNT5A-mediated GSC differentiation and invasive growth in disease recurrence. Thus, the PAX6/DLX5-WNT5A axis governs the diffuse spread of glioma cells throughout the brain parenchyma, contributing to the lethality of GBM.

Keywords: cancer stem cell differentiation; glioblastoma; recurrence; tumor microenvironment.

Figure 1. Overexpression of p53DN and myr-AKT Generates Malignant Glioma and Upregulates EC Signaling Pathway

(A) Immunoblot analysis of overexpressed oncogenes in hNSCs. (B) Soft agar colony formation of hNSCs expressing p53DN, p53DN/myr-AKT (p53DN-AKT). Error bars represent SD of triplicate wells. **p < 0.01. Representative images are shown. (C) Kaplan-Meier survival analysis for oncogenic transformation of hNSC in vivo. (D) Representative H&E image of intracranial tumor derived from p53DN-AKT-hNSCs; scale bars, 1 mm. (E) Representative H&E image of tumor sections with necrotic area (N) and microvascular hyperplasia (black arrow). Scale bars, 50 μm. (F) IHC staining of tumors with the indicated antibodies. Scale bars, 50 μm. (G) Top ten signaling pathways related to hNSC oncogenic transformation were identified by GSEA analysis based on gene expression profiles of hNSCs and their derivative cells. The normalized enrichment scores (ES) and the log transformed p values are shown. (H) GSEA enrichment plots of genes ranked based on oncogenic transformation versus EC signaling pathway. (I) Heatmap of histone landscape of gene transcriptional start sites (TSSs) within ±2 kb and of Log2-ratio of these gene expression levels in hNSCs and iGSCs. See also Figure S1 and Table S1.

Figure 2. Activation of AKT Pathway Induces Differentiation of GSCs into ECs

(A) FACS analysis of hNSCs, p53DN-transduced hNSCs, and p53DN-AKT-hNSCs based on CD133 and CD144 expression. (B) Fold change of percentage of CD133⁺/CD144⁺ cells by FACS analysis in p53DN-AKT-hNSCs under treatment with rapamycin (RAPA, 50 nM) for 72 hr. (C) qRT-PCR for indicated EC markers expression in two sorted subpopulations from p53DN-AKT-hNSCs. (D) IF analysis of sorted CD133⁺/CD144⁺ from p53DN-AKT-hNSCs cultured under NSC or EC media for 5 days for EC markers expression and DiI-AcLDL uptake. Scale bar, 40 μm. (E) Tubular networks formation of sorted CD133⁺/CD144⁺ and CD133⁺/CD144⁻ cells from p53DN-AKT-hNSCs cultured on Matrigel in EC media with/without RAPA (50 nM) treatment. Scale bar, 100 μm. (F) Immunoblot analysis of AKT/mTOR pathway activation in patient-derived GSCs. (G) FACS analysis of CD133⁺/CD144⁺ cells in the indicated GSCs. (H) FACS analysis of CD133⁺/CD144⁺ cells in the indicated GSCs with myr-AKT overexpression. (I) FACS analysis of CD133⁺/CD144⁺ cells in the indicated GSCs treated with RAPA (50 nM) for 72 hr. Error bars represent SD of the mean of two (C and G) or three (B, H, and I) independent experiments. **p < 0.01. See also Figure S2 and Table S2.

Figure 3. AKT-Driven WNT5A Upregulation in GdECs Differentiation of hNSCs

(A) FACS analysis for the percentage of CD133⁺/CD144⁺ cells in7 days post-infection p53DN-hNSCs cells by lentivirus carrying the indicated genes individually. (B) Quantitation of the percentage of CD133⁺/CD144⁺ cells in (A) from four independent experiments. (C) Matrigel tubular network formation of the sorted CD133⁺/CD144⁺ cells from p53DN-AKT-hNSCs with infection by lentivirus carrying pooled short hairpins (minimum three shRNAs) targeting each indicated gene. (D) Quantitation of the number of tubular networks branch points in (C) (n = 5). (E) FACS analysis of CD133⁺/CD144⁺ cells in p53DN-hNSCs overexpressing myr-AKT or WNT5A with BOX5 treatment (50 μM) for 72 hr. (n = 3). (F) Representative images for the tubular network of sorted CD133⁺/CD144⁺ cells from p53DN-AKT-hNSCs with BOX5 treatment. Scale bar, 100 μm. (G) Number of branch points calculated in (F) (n = 5). Error bars represent SD of the mean; **p < 0.01. See also Figure S3 and Tables S1 and S3.

Figure 4. Transcriptional Activation of WNT5A by PAX6 and DLX5

(A) ChIP-seq analysis of chromatin status for WNT5A locus around TSS in hNSC and iGSC. (B) PAX6 and DLX5 binding motifs in WNT5A regulatory regions. (C) Chromatin modification changes from hNSC to iGSC for PAX6. The peak of H3K27me in iGSC is highlighted in sky blue color. (D) Binding of PAX6 in WNT5A regulatory regions in hNSC by ChIP-PCR. Beta-actin locus (ACTB_exon) was used as the negative control (n = 3). (E) Chromatin modification changes from hNSC to iGSC in DLX5-DLX6 locus. (F) Binding of DLX5 in WNT5A regulatory regions by ChIP-PCR. PAX2 was used as the control for non-specific binding (n = 3). (G) WNT5A expression by qRT-PCR analysis in GSCs and iGSC-overexpressing PAX6 (n = 3). (H) WNT5A expression by qRT-PCR analysis in GSCs and iGSC-overexpressing DLX5 (n = 3). Error bars represent SD of the mean; *p < 0.05 and **p < 0.01. See also Figure S4.

Figure 5. WNT5A-Mediated Endothelial Lineage Differentiation in Tumor Neovascularization and Satellite Lesion Formation

(A) Representative images for the hemorrhage lesion in mouse brain that received injection of TS543-overexpressing WNT5A (WNT5A OE) versus control (Vector). H&E and IHC analyses of tumor sections show the microvascular hyperplasia (black arrows) and expression of CD34 and WNT5A. Scale bar, 50 μm. (B) Representative images for the satellite lesions in peritumoral areas. Scale bar, 200 μm. (C) Representative images for GdECs (yellow arrows) identified by co-staining with TRA-1-85 and CD34 in intratumoral and peritumoral areas. Scale bar, 25 μm. (D) Quantitation of TRA-1-85⁺/CD34⁺ cells using Vectra software system (n = 3 tumors). (E) High magnification of rectangle area in (C). Scale bar, 10 μm. (F) IHC staining of CD34 in intracranial tumors derived frompCD144-GFP infected WNT5A-TS543following GCV treatment. Representative images of low (scale bar, 100 μm) and high (scale bar, 50 μm) magnification. (G) Dotplots for quantitation of MVD in tumors with/without GCV treatment (n = 4 tumors, five fields per tumor). (H) Representative images for tumor appearance (left, scale bar, 2,000 μm) and peritumoral satellite lesions (right, scale bar, 200 μm). See also Figure S5.

Figure 6. Recruitment of Host ECs by WNT5A-Mediated GdECs Contributes to GSCs Self-Renewal and Proliferation

(A) Representative images of IF analysis for GdECs (green arrows), compared with tumor cells (red arrows), are in close proximity to mouse ECs (white arrow) in tumor sections. Scale bar, 10 μm. (B) Dotplots show the distance from mouse ECs to the nearest tumor cells and GdECs, respectively (n ≥ 15). (C) Illustration of the transwell system to measure EC recruitment. (D) Fluorescence intensity shows HBMECs recruitment after co-culture with GdECs for 24 hr (n ≥ 3). (E) qRT-PCR for CD144 and WNT5A mRNA levels in sorted pCD144-GFP^– and pCD144-GFP⁺ from TS543-WNT5A and TS603 (n = 3). (F) Fluorescence intensity shows HBMECs recruitment after co-culture with NSC media containing rWNT5A (0.5 μg/ml) or rWNT3A (0.05 μg/ml) (n = 3). (G) Representative images of GdECs (green arrows) and mouse ECs (white arrows) in variously sized satellite lesions. Scale bar, 20 μm. (H) Neurosphere formation of TS543 or TS603 co-cultured with GdECs and HBMECs (n = 3). Cartoon depicting the experimental approach. Error bars represent SD of the mean; *p < 0.05 and **p < 0.01. See also Figure S6.

Figure 7. Correlation of WNT5A-Mediated GdEC with Peritumoral Satellite Lesion and Tumor Recurrence in GBM Patients

(A) Representative images of GdECs (yellow arrows) defined using indicated EC and GSC markers. White arrows denote host ECs. Scale bar, 20 μm. (B) Representative images with IHC double-staining and cell segmentation obtained from Caliper InForm analysis software show the close proximity of GdEC (SOX2⁺/CD31⁺, yellow) and host ECs (SOX2⁻/CD31⁺, green) compared with GSCs (SOX2⁺/CD31⁻, red) in tumor sections. SOX2⁻/CD31⁻ cells are marked in blue color. Scale bar, 20 μm. (C) Boxplot of distances from host ECs to the nearest GSCs and GdECs, respectively (n = 300). (D) The correlation between WNT5A mRNA expression and GdEC signature score. n = 364 (IDHwt GBMs); mRNA expression was normalized across genes. (E) Representative image of H&E staining for intratumoral and peritumoral regions (black dashed line) of GBM patient's sample. Black arrows denote peritumoral satellite. Scale bar, 200 μm. (F) Representative images for GdECs (black arrows) and host ECs (red arrows) in variously sized satellite lesions in IHC double-staining tumor sections. Scale bar, 25 μm. (G) Fourteen patients' primary tumors were divided by WNT5A staining index into two groups (low and high). Tumor sections with peritumoral satellite lesions (more than ten) were counted as the highest score. *p = 0.04 by the log-rank test for PFS between two groups, HR = 3.45 (high versus low). (H) Comparison of WNT5A mRNA expression between nine pairs of intratumor and peritumor regions from GBM patients. Each dot in the scatterplot represents a pair. Boxplot summarizes the distribution of WNT5A expression in nine intratumor and peritumor regions, respectively. (I) TCGA GBMs (IDHwt, n = 228) were used for PFS analysis. Red and blue lines show survival curves of top 20% of GBMs with highest and lowest WNT5A mRNA expression, respectively. (J) Representative images for WNT5A (brown) and CD31 (red) staining of paired primary/recurrent tumors from one GBM patient. Scale bar, 25 μm. (K) Unbiased quantification of GdEC frequency in primary and recurrent GBMs (n = 150). (L) Correlation between WNT5A expression and GdEC signature scores in recurrent GBMs. Small boxplot panel shows all 81 pairs while the big boxplot panel shows the majority of samples. (M) Association of differences of WNT5A mRNA expression and GdEC signature score between 81 matched primary/recurrent GBMs pairs. Each circle in the scatterplot represents a GBM pair; mRNA expression was normalized across genes. See also Figure S7 and Tables S2, S4, S5, and S6.