The dystroglycan receptor maintains glioma stem cells in the vascular niche

Abstract

Glioblastomas (GBMs) are malignant central nervous system (CNS) neoplasms with a very poor prognosis. They display cellular hierarchies containing self-renewing tumourigenic glioma stem cells (GSCs) in a complex heterogeneous microenvironment. One proposed GSC niche is the extracellular matrix (ECM)-rich perivascular bed of the tumour. Here, we report that the ECM binding dystroglycan (DG) receptor is expressed and functionally glycosylated on GSCs residing in the perivascular niche. Glycosylated αDG is highly expressed and functional on the most aggressive mesenchymal-like (MES-like) GBM tumour compartment. Furthermore, we found that DG acts to maintain an MES-like state via tight control of MAPK activation. Antibody-based blockade of αDG induces robust ERK-mediated differentiation leading to reduced GSC potential. DG was shown to be required for tumour initiation in MES-like GBM, with constitutive loss significantly delaying or preventing tumourigenic potential in-vivo. These findings reveal a central role of the DG receptor, not only as a structural element, but also as a critical factor promoting MES-like GBM and the maintenance of GSCs residing in the perivascular niche.

Keywords: Dystroglycan (DG); EphA3; Glioblastoma (GBM); Glioma stem cell (GSC) commitment; Integrin-α6; MAPK signalling; MES-like GBM; Perivascular niche.

Fig. 1

Elevated Dystroglycan Correlates with GBM Patient Outcome and αDG is Abundantly Glycosylated in GBM. aDAG1 expression was correlated with GBM patient survival using the Rembrandt (n = 523) and TCGA (n = 453) databases. b QPCR analysis of DAG1, ITGA6, EPHA2 and EPHA3 mRNA expression in GBM tissue specimens (n = 28). c Flow cytometric analysis for αDG glycosylation (IIH6 mAb) was performed on primary GBM cell lines grown as serum-free GNS cultures, compared to isotype control. See also Online Research 1d for full analysis. d αDG glycosylation was assessed by western blot in four primary GBM cell lines. e αDG glycosylation was assessed by western blot following cell fractionation to compare cytoplasmic versus membrane localisation. β-actin was used as a loading control. f Flow cytometric analysis was performed for αDG glycosylation (IIH6 mAb) and EphA3 (IIIA4 mAb) in 10 early passage primary GBM cultures, mean channel fluorescence (mcf) was used to determine the correlation coefficient between EphA3 and glycosylated αDG (r = 0.899). GBM subtypes: MES mesenchymal, PN proneural, CL classical

Fig. 2

Glycosylated αDG is expressed in the vascular niche and discretely on Mesenchymal-like glioma tissue. a H&E section of a Gliosarcoma (GS) patient specimen showing characteristic biphasic gliomatous (glial-like) and sarcomatous (mesenchymal-like) tumour elements. b IHC analysis of a GS specimen to identify the mesenchymal-like, vimentin⁺, tumour element. c IHC analysis of sequential (#1–6) GS tissue sections was performed to assess expression patterns of EphA2, EphA3, αDG, βDG, CD31 and GFAP. d IF dual staining of a GBM specimen showing localisation of αDG expression (IIH6—green) surrounding (CD31⁺—red) tumour blood vessels, nuclei DAPI—blue). See also Online Resource 2 for additional IHC and IF analysis in GBM tissue specimens

Fig. 3

αDG Interacts with EphA2 and EphA3 Receptors and is Expressed on GSCs. a IF staining was performed to compare membrane localisation of glycosylated αDG (IIH6 mAb—red), compared to either EphA2 (1F7 mAb—green) or EphA3 (IIIA4 mAb—green), in four primary GBM cultures. b Immunoprecipitation (IP) for EphA3 was performed from 1 mg of lysate from four early passage primary GBM cultures to compare membrane association of EphA2, EphA3 and glycosylated αDG. Protein G only was used as a control. c Amnis flow cytometric analysis was performed on a dissociated GBM patient tissue specimen (n = 1) to assess membrane localisation of glycosylated αDG, with known GSC markers (CD15, CD133, CD49f, EphA2 and EphA3). d Amnis analysis and flow cytometry was performed on primary GBM cultures to assess membrane localisation of glycosylated αDG, with known GSC markers. e αDG^high versus αDG^low populations were isolated from WK1 cells using FACS and cell morphology and sphere number assessed 7 days post-sort (*p < 0.05). GBM neuropshere differentiation was induced using 2% FBS, 3 days post-differentiation glycosylated αDG expression was assessed by flow cytometry and GSC and diff marker expression assessed by QPCR and compared to undifferentiated cells, see Online Resource 3 for complete analysis. All data presented as the mean ± SD of three independent experiments

Fig. 4

αDG Blockade Induces GSC Differentiation. a Four primary GBM neurosphere cultures treated with the αDG blocking mAb (IIH6, 50 µg/ml) and an equivalent IgM control (50 µg/ml). Bright field images 24 h post-treatment showing loss of neurosphere formation and cell adherence. b IF staining was performed 48 h post IIH6 mAb treatment for the differentiation markers (GFAP-red, βIII-tubulin—green, myelin basic protein (MBP)—red and the nuclear counter-stain DAPI-blue). WK1 data shown see also Online Resource 4b for complete analysis. c IncuCyte analysis was performed to quantitate cell adhesion in real time following IIH6 treatment-blue compared to an equivalent IgM control-red for 7 days (*p < 0.05). d Cell proliferation was assessed by direct cell counting using a haemocytometer 7 days post IIH6 treatment compared to IgM control and untreated cells (*p < 0.01). e ApoTox-Glo Triplex assay was used to assess caspase3/7 activity and cell viability 48 h post IIH6 treatment. WK1 data shown see also Online Resource 4c for complete analysis. f Neurosphere formation was assessed in WK1 cells following IIH6 withdrawal or re-addition following 7 days initial treatment. Bright field images at 2 weeks showing neurosphere re-formation following IIH6 withdrawal and cell death following IIH6 rechallenge. Data presented as the mean ± SEM of three independent experiments

Fig. 5

αDG Controls ERK Signalling to Regulate GSCs and Promote an MES-like GBM State. a ERK1/2 phosphorylation (pERK1/2) was assessed by western blot in WK1 GBM neurospheres following IIH6 (50 µg/ml) treatment for 24 h, compared an equivalent IgM control mAb (50 µg/ml) and untreated cells. β-Actin was used as a loading control. b IF staining was performed 24 h post IIH6 mAb treatment for pERK1/2—green and αDG—red to assess nuclear versus cytoplasmic localisation, 2% FBS was used as positive differentiation control. c IHC analysis of sequential (#1–4) GS tissue sections was performed to assess discrete expression patterns of αDG, EphA3, GFAP, pERK and Ki67 (non-sequential). d QPCR analysis of integrin α6A, integrin α6B, E-cadherin, N-cadherin and Slug mRNA expression in GBM specimens (n = 28). e Correlation coefficient analysis of mRNA expression of the mesenchymal markers (Slug, Snail, Twist, Vimentin and N-cadherin) and the GSC markers (EphA3, CD15, CD133, CD49f and SOX2) in 6 primary early passage GBM GNS cultures. f QPCR analysis of VEGF-A, BMI-1 and ESRP1 mRNA expression in GBM specimens (n = 28). g ERK1/2 phosphorylation (pERK1/2) was assessed by western blot in WK1 GBM neurospheres following GoH3 (10 µg/ml) or IIH6 (50 µg/ml) either alone or in combination compared an equivalent control mAb combination IgG2a (10 µg/ml)/IgM(50 µg/ml), β-actin was used as a loading control. h Cell proliferation was assessed by direct cell counts using a haemocytometer 7 days post IIH6 treatment alone or combined with GoH3. Data presented as the mean ± SD of three independent experiments, *p < 0.05

Fig. 6

DAG1 down regulation delays or prevents GBM formation in-vivo.a IF staining and QPCR was used to assess DAG1 mRNA levels following shRNA mediated KD in primary GBM cell lines. b Cell proliferation was assessed by direct cell counts using a haemocytometer following stable DAG1 KD. Data presented as the mean ± SD of three independent experiments, *p < 0.05. c Bright field images of WK1 and JK2 neurospheres 2 week post-DAG1 shRNA KD compared to control shRNA transfected cells. See also Online Resource 6a for SJH1 images. d Kaplan Meier analysis showing a significant p < 0.05 increase in overall survival following orthotopic intracranial engraftment of 1 × 10⁵DAG1 shRNA versus control shRNA cells into the right striatum of NOD-SCID mice. WK1 n = 4 animals per arm, JK2 n = 7 animals per arm. e Representative H&E coronal sections from DAG1 shRNA versus control shRNA engrafted animal following euthanasia from either illness or tumour burden. See Online Resource 6d for WK1 H&E images

Fig. 7

The dystroglycan complex supresses MAPK activation to regulate glioma Stem cell commitment. DG, when bound to laminin, cooperates with EphA3 and integrin α6B in the vascular niche to maintain a GSC mesenchymal phenotype. Functionally, DG and integrin α6B limit sustained ERK activation preventing GSC differentiation. Blockade of αDG glycosylation sites, using the IIH6 mAb, induced sustained ERK1/2 activation, leading to translocation of ERKs to the nucleus followed by GSC differentiation and reduced GBM aggressiveness