Mesenchymal glioma stem cells trigger vasectasia—distinct neovascularization process stimulated by extracellular vesicles carrying EGFR

Targeting neovascularization in glioblastoma (GBM) is hampered by poor understanding of the underlying mechanisms and unclear linkages to tumour molecular landscapes. Here we report that different molecular subtypes of human glioma stem cells (GSC) trigger distinct endothelial responses involving either angiogenic or circumferential vascular growth (vasectasia). The latter process is selectively triggered by mesenchymal (but not proneural) GSCs and is mediated by a subset of extracellular vesicles (EVs) able to transfer EGFR/EGFRvIII transcript to endothelial cells. Inhibition of the expression and phosphorylation of EGFR in endothelial cells, either pharmacologically (Dacomitinib) or genetically (gene editing), abolishes their EV responses in vitro and disrupts vasectasia in vivo. Therapeutic inhibition of EGFR markedly extends anticancer effects of VEGF blockade in mice, coupled with abrogation of vasectasia and prolonged survival. Thus, vasectasia driven by intercellular transfer of oncogenic EGFR may represent a new therapeutic target in a subset of GBMs.

Supplementary Figure 6.BME plug assay for assessment of blood vessel stimulating activities of tumour cell derived extracellular vesicles in the absence of cancer cells.EVs were isolated from cultures of EGFR-proficient (WT) and -deficient (KO) glioma stem cells (GSC83) and added to BME matrix before subcutaneous injection into immune deficient NSG mice.Recombinant VEGF (25 ng/mL) was addad, as a positive control.Ten days post injection BME plugs were removed and photographed (n=3 independent experiments).GSC83 xenografts spatial sequencingtop differentially expressed genes (DEGs) Large-WT Small-WT Large-KO Small-KO Large-CTRL Small-CTRL Interferon response-related genes (Normalized to PECAM-1)

Nanoparticle tracking analysis
EV size and quantity were analyzed using NS500 (Nanosight; Malvern Panalytical, Malvern, UK) nanoparticle tracking (NTA) instrument.NanoSight technology uses light scattering to visualize particles in the range of 100 nm to 2 μm, records their movement in 30 second video files, tracks individual particles and calculates concentration and size based on Brownian motions.NTA was performed on crude culture media and purified vesicles, as indicated.For culture media, 500 μL aliquot was collected and centrifuged at 400g for 10m to remove cells, then the sample was diluted with D-PBS to reach optimal loading concentration of 10 7 to 10 9 particles per mL.For extracted EVs, the vesicles were diluted 1:200 in D-PBS.Three recordings of 30 seconds were taken for each sample and the analysis was performed as described earlier 2 .
Amersham ECL Western Blotting Detection Kit (RPN2108 GE Healthcare) was used for Detection of Chemiluminescence and band visualization using ChemiDoc MP system (Biorad).

Cell growth/survival assays (MTS Assay)
Cell titer 96 (Promega # 43580) kit was used to measure in vitro cell growth/viability in the presence of Dacomitinib or EGFR knockout.As indicated, 7 x10 3 GSC cells/well were seeded in 96 well plates in full growth media for 24h.The following day the cells were washed and treated with 2.5µM Dacomitinib in DMEM containing 1%FBS.For EGFR-KO and EGFR-WT analysis the cells were left in full growth media for the duration of the assay.The absorbance at 490 nm was read at time intervals indicated, and the signal reflective of viable cell numbers was assessed for up to six days following manufacturer's instruction.

Transwell migration assay
Gelatin (0.1%) coated 8.0μm trans-well inserts (Thermo Fisher Scientific) were placed in 24-well plate and HUVEC (2 x 10 3 ) cells were plated into the inserts on day 0 of the experiment with full media.The following day, media was removed, HUVEC cells were washed in PBS twice and starvation media containing 1% EV-depleted FBS 3 was added to the cells and the lower part of the well.Conditioned media (1:1), EV-depleted supernatant (1:1), or reconstituted EVs (30ug/mL), from each cell line, versus control buffer (PBS) containing no EVs were added onto the HUVEC cells to stimulate their migration.After the incubation of 3 days, inserts containing cells were fixed with 3.7% formaldehyde for 5 min and washed twice with PBS.This step was followed by staining with 0.5% crystal violet solution (0.5g crystal violet, 20ml methanol, 80ml water; mixed and filtered through a 0.45μm filter) for 10 min and washed with PBS until no excess stain was left on the inserts.Finally, the non-migrated cells (inside the inserts) were removed manually by gently swabbing the inside surface of each insert using cotton swabs without damaging the insert membrane.Finally, the inserts were examined under the light microscope to evaluate the number of cells that have migrated through the membrane to the bottom surface of the insert.Quantification was performed using FIJI software.

Immunofluorescent staining (IF)
Tumor tissues were preserved in 4% PFA immediately after resection from the mice.They were then run through a series of automated processing steps executed in a Leica TP 1050 tissue processor according to user manual.The resulting paraffin embedded blocks where sectioned using American Optical microtome into 4 μm thick tissue sections, which were placed on permanent positive charge microscope slides and stored until used.Tumor tissue containing slides where dewaxed in Xylene, and then re-hydrated in a series of alcohol washes (95% ethanol ->50%ethanol).
Slides were then re-hydrated and antigen retrieval was performed in 0.01 M citrate buffer (pH 6.0)We used 2x HBS mixed with a solution containing 2M Calcium and the plasmid mixture mentioned above for transfection procedure.Transfected cells were then incubated over-night (16-18hr) at 37C, 5% CO2 and on the next day the conditioned media was collected, and debris was spun down at 1,500 x g for 10 min (4 o C), Finally, the pellets of viral particles were collected at 22,000 rpm for 2 hrs.These pellets were then re-suspended in 50 μL of PBS.pSMAL lentiviral vectors were used to transduce MES-GSC throughout our experiments.
After 16 h, the cells were fed with additional appropriate growth media, and transduction efficiency was measured using flow cytometry for GFP or RFP, as appropriate.

Intracranial injections
The NSG (NOD scid IL2Rgamma-/-) transgenic mice were injected intracranially with GSCs (25,000 cells/μL with total volume of 2μl) using a Stoelting Stereotaxic Injector at pre-determined coordinates (2.5-1.5-3.0) of bregma and sagittal suture, as described 4  15μg/ml) substrate for 20-30 minutes.These measurements were restricted to mice that had been injected with Luciferase expressing cells and used to visualise the extend and dynamics of tumour formation.Frequent monitoring and humane clinical endpoints were strictly observed.

BME plug vascular growth assay
Cold liquid growth factor-reduced Cultrex basement membrane extract (BME) solution (3433-010-R1 R&D system) was mixed with 100 μg of appropriate GSC EVs or VEGF at indicated concentrations, injected subcutaneously into C57BL/6 mice and allowed to solidify to form a palpable pellet.The injections were all placed at the same location, using a major mammary vessel running along the flank of each mouse and visible through the shaved skin.From the last rib we followed the vessel until its first branching point, which we labeled with a marker and used as the injection point.Pellets were allowed to become vascularized and were collected on day 21 post injection, photographed, imaged under light microscope and placed in sucrose for cryopreservation and histology.

Aortic ring endothelial outgrowth assay
Aortas of 4-week-old C57BL/6 mice were isolated, cleaned of the surrounding tissue and cut into 1 mm long segments (rings).Next, the rings were cultured in growth factor-reduced Cultrex BME (3433-010-R1 R&D system) polymerized at 37 °C.The rings were observed every-day until sprout-like vascular protrusions started to emerge.At that point rings were washed 4 times with PBS and placed in starvation media containing 1% FBS and supplemented with either 30μg/mL of EVs, VEGF, or vehicle.Vascular patterns (angiogenesis and other forms of vascular growth) were assessed at 7 days by using an inverted microscope platform.The number of sprout-like endothelial outgrowths were quantified using Images, which were analyzed using the Fiji distribution of ImageJ (PMID: 22743772) with Angiogenesis Analyzer plugin.

VEGF Elisa
For the detection of VEGF secreted from GSC cells, either as a soluble factor in the supernatant or released in EVs we employed ELISA kit purchased from R&D Systems (#RRV00) and used according to manufacturer's protocol.In order to account for both surface and luminal VEGF, EVs were lysed in RIPA buffer and RIPA was included in, and subtracted from, the assay readings.
The readings were collected from three independent samples for each experimental condition, normalized to the total volume and read at 2 times dilutions against a standard curve.

Phosphorylated Receptor Tyrosine Kinase Antibody Array
To demonstrate the intracellular signaling mechanisms elicited by EVs in HUVEC or HBEC-5i cells, cultured endothelial cells were stimulated with indicated EVs at 30 μg/mL.For this purpose GSC-derived EVs were combined with endothelial cells and incubated for 6 days.On the last day, the cells were collected, lysed in RIPA buffer and the relative expression of phosphorylated kinases were analysed using commercial anti-phospho-protein antibody arrays covering 49 different receptor tyrosine kinases (RTKs) and 26 downstream kinases, including 9 mitogen activated protein kinases (MAPKs).The respective kits included Human Phospho-RTK Array kit and the Human Phospho-MAPK Array kit (both from R&D Systems, Minneapolis, MN, USA), respectively, used according to the manufacturer's protocols.Each array was incubated with 250 μg of protein lysate and the levels of phosphorylation were detected by chemiluminescence and quantified using Fiji software.Pixel densities of duplicate spots were averaged, and the value of background was subtracted.

Immunoprecipitation of EGFR-positive EVs
For immunoprecipitation we used Dynabeads™ Protein G, which are uniform, 2.8 µm in diameter, superparamagnetic beads with recombinant Protein G (~17 kDa) covalently coupled to their surfaces.After washing with PBS twice the beads were incubated overnight with rabbit anti-EGFR antibody (4267 Cell signaling).The day after the antibody was removed, beads were washed three times in PBS and incubated overnight with intact EVs (30ug).Using a magnet, EVs enriched with EGFR proteins were separated from the flow through fraction, which contained EVs not carrying EGFR proteins.The EGFR-positive EVs and EGFR-negative EVs were then lysed either in RIPA buffer or in Lysis buffer for RNA extraction (as mentioned above).Protein validation was

Supplementary Figure 1 . 2 .
Vascular activities of soluble and vesicular components of proneural and mesenchymal glioma stem cell secretomes.a. Concentration-dependent stimulation of endothelial cell (HUVEC) migration by recombinant human VEGF (rhVEGF).After 3 days post stimulation cells were fixed and stained with crystal violet to assess the number of cells migrated through the pores using FIJI software (n=2 independent experiments, two-tailed paired t test; P= 0,012 -0,013 -0,0077 -0,00093 and 0,0053); b.Aortic ring assay -endothelial outgrowth responses induced by the whole conditioned media, EV-depleted supernatant and EVs of GSC83 cells (3 days post treatment); c-d.HUVEC or HBMVEC transwell migration assay in the presence of whole conditioned media (left), supernatant (middle) or EVs (right) from indicated proneural (PN), or mesenchymal (MES) glioma stem cells (GSCs).Endothelial cells were seeded in growth factors-free media on 0.1% gelatin-coated transwells filters for 24h, then treated as indicated, and 3 days later fixed, stained and enumerated (FIJI software) against rhVEGF positive control (25 ng/mL).Data were presented as means ± SD (n=3 and 4 independent experiments; two-tailed paired t test; * p < 0.05, **p < 0.01, ***p < 0.001 for treated groups versus controls.Scale bars are 250 μm.Mesenchymal glioma stem cells release extracellular vesicles that transfer to endothelial cells the expression of EGFR/EGFRvIII oncoproteins.a.Quantification of the human phospho-RTK signal from the antibody array analysis performed on immortalized human brain endothelial cells (HBEC-5i) treated with EVs from either proneural (PN)-or mesenchymal (MES)-GSC for 7 days.MES-GSC (GSC83) EVs, but not PN-GSC EVs (GSC157) trigger the expression of phosphorylated EGFR in endothelial EV recipients; data presented as means ± SD (n=3 and 4 independent experiments, respectively; two-tailed paired t test; ***p < 0.001 for MES-GSC EV-treated group versus no-EV-treated group); b.Primary endothelial cells (HUVEC) incubated with EVs from either PN-or MES-GSC EVs were immunoblotted for EGFR, phospho-EGFR (pEGFR) and beta-actin control.Only MES-GSC EVs triggered EGFR expression/phosphorylation in endothelial cells, detectable up to 6 days post treatment (n=3 independent experiments); c.EVs from MES (GSC83) but not PN (GSC157) cells transfer the expression of EGFR protein to human brain microvascular endothelial cells (HBMVECs; n=3 independent experiments); d.Both EGFR and mutant EGFRvIII are detectable in HBEC-5i endothelial cells following incubation with EVs from GSC83 cells (n=3 independent experiments); e. EV uptake and transfer of GSC subtype specific cargoes: CD133 transfer to endothelial cells by EVs from PN GSC donors (Flow cytometry; n=3 independent experiments).

Supplementary Figure 3 .Supplementary Figure 5 .
Characterization of glioma stem cell derived extracellular vesicles.a. Mesenchymal glioma stem cell (GSC83)derived EVs transfer EGFRvIII mRNA to human primary endothelial cells (HUVEC; signal detected up to 6 days post treatment; n=3 independent experiments); b.Nanoparticle tracking (NTA) analysis of EVs released into cell culture media following treatment of MES-GSCs (GSC83, GSC1005) with different concentrations of GW4869 inhibitor (5 -10 μM) for 24h; the levels of vesiculation remained relatively unchanged.(n=3 independent experiments).mRNA to HUVEC cells Supplementary Figure 4. Pharmacological targeting of EGFR in EV donor cells.a. Effects of different Dacomitinib (DAC) concentrations on EGFR phosphorylation in mesenchymal glioma stem cells.GSC83 cultures were treated with various concentrations of DAC (0.06 μM to 5 μM) for 24h after which the cells were analyzed by western blot.DAC exposure led to near complete loss of EGFR phosphorylation at 0.250 μM and above.(n=3 independent experiments).b.DAC inhibition of endothelial cell migration stimulated by EGFR-carrying EVs.Endothelial cells were seeded in growth factors-free media on 0.1% gelatin-coated transwells.After 24h they were treated with MES EVs, or VEGF and 6 hours later with 2.5 μM of Dacomitinib.After a period of 3 days, cells were fixed and stained with crystal violet to assess the number of cells migrated (n=3 independent experiments).Genetic targeting of EGFR in EV donor cells.a. Sanger electropherograms of short PCR products around the gRNA regions in Cas9-gRNA transduced single clones of indicated MSE-GSC lines.Genetic analysis of EGFR indicates the insertion of a single adenine for 83-KO-19, 83-KO-27 and 1005-KO-11, and for MES-GSC1005-KO-14 a 10-nucleotide deletion.All the CRISPR/Cas9-mediated genome editing induced a disruption in gene expression and function; b.Validation of EGFR knockout by western blotting.(n=3 independent experiments); c.EGFR depletion reduced the ability of GSC EVs to trigger endothelial cell migration.Endothelial cells were treated with 30 μg/mL of EVs derived from cells deficient (EGFR-KO) or proficient (EGFR-WT) for EGFR.After 3 days cells were fixed, stained with crystal violet and imaged.(n=3 independent experiments; two-tailed paired t test; Significance of difference for GSC83-EVs: P=0,00036 and 0,00014; for GSC1005-EVs: P=9,84E-05 and 0,00037); d.EGFR depletion reduced the ability of EVs to trigger endothelial cell outgrowths from aortic rings.Isolated rings were treated with 30 μg/mL of EVs obtained from GSCs with EGFR-KO or EGFR-WT status.For both assays 25 ng/mL of rhVEGF was used as a positive control (n=3 independent experiments; data were presented as means ± SD; **p < 0.01, ***p < 0.001 for treated group versus controls; Scale bar in red is 250 μm; Scale bar in blue is 0.5 mm).

Supplementary Figure 7 .
Disruption of EGFR gene in mesenchymal glioma stem cells results in diminished tumour aggressiveness and abrogation of vasectasia-like vascular patterns.a. Impact of EGFR gene editing on GSC1005 xenografts; intracranial tumour growth in mice was measured by symptom free survival -comparison of two independent EGFR-KO clones (1005-KO-11 and 1005-KO-14) and two controls (1005 WT and 1005 OR56A1; n=5 independent experiments; two-tailed paired t test; P= 0.000103 and 0.000010); b Representative images of immunofluorescence for CD31 reveals differential vascular patterns between tumours driven by MES GSCs with different EGFR status: EGFR-WT or MES GSC EGFR-KO; n=5 independent experiments); c.Quantification of vessel size distribution through CD31 staining of endothelial cells (n=5 independent experiments; two-tailed paired t test; P= 2,32 -05 and 5,23 -05 ); d.Quantification of microvascular density using CD31 staining (n=5 independent experiments; two-tailed paired t test; P= 7,40 -16 and1,30 -16 ).Microvascular density was expressed as vessel numbers per high power field (hpf; scale bars are 50 µm).Supplementary Figure 8. Rescue of the agressive phenotype and vasectesia-like vascular patterns upon selective re-expression of EGFRvIII in mesenchymal GSCs with disrupted EGFR gene.a. EGFRvIII expression vector; vascular properties of GSC83 and GSC1005 intracranial xenografts with disrupted EGFR, and either untransfected (83-KO-19), transfected with empty vector (1005-KO-14-pgLNCX), or reconstituted for selective EGFRvIII expression (83-KO-19-EGFR+; 1005-KO-14-EGFR+); b.Quantification of vessel size distribution through CD31 staining of endothelial cells (n=5 independent experiments; two-tailed paired t test P= 1,19 -06 and 7,67 -11 ); c.Quantification of microvascular density using CD31 staining (n=5 independent experiments; two-tailed paired t test; P= 1,68 -05 and 0,00037); Microvascular density was expressed as vessel numbers per high power field (hpf).d.Kaplan-Meier survival curves of mice bearing EGFR-WT or EGFR-KO MES GSC-derived tumors.Survival of tumor bearing mice Supplementary Figure 9. Spatial gene expression profiling (GeoMX) in GSC83 intracranial tumour regions containing large (extended) and small (capillary) vascular structures.a. Regions of interest (ROIs) around vessels highlighted with anti-CD31 fluorescent immunostaining; b.Characteristics of sequencing reads; c.Clustering of transcriptomes in ROIs; d.Differentially expressed genes (DEGs) in indicated ROIs containing large or small vessels, as a function of either wild type (WT) or knock-out (KO) EGFR status; e. Volcano plot of statistical differences in DEG between ROIs containing capillary (small -KO tumours) and vasectasia-type (large -WT tumours) vasculatures; f.Comparison between trascriptomes in ROIs containing small or large vascular structures in WT or KO tumours, versus corresponding vessel calibers in control mouse brains.
GSC83 xenografts spatial sequencing -DEGs as a function of vessel diameter and EGFR status Reads across whole study .Spatial expression profiling (GeoMX) of gene subsets in vascular regions of brain tumours driven by GSCs that are either proficient or deficient for EGFR/EGFRvIII.a. GSEA analysis of pathways enriched in vascular regions (ROIs) containing large vessels (83 WT tumours -EGFR/EGFRvIII expressors), or small/capillary vessels (83 KO tumours -EGFR/EGFRvIII non-expressors) normalized to normal brain vessels; NES = normalized enrichment score); b.Differentially expressed genes (DEGs) related to interferon response within indicated ROI; c. DEGs related to angiogenesis within indicated ROI; d.DEGs related to cellular proliferation including expression associated with G1/S or G2/M phases of cell cycle.

a
of GeoMx mRNA profiles between glioma xenograft-associated vascular regions and regions related to vascular structures of corresponding sizes in normal mouse brain.Transcriptomes of endothelial cells in key categories of normal blood vessels extracted from the ATLAS of normal mouse brain vasculature (Vanlandewijck et al 2018) were compared to the expression levels of the respective genes in vascular ROIs containing large or small vessels in EGFR-proficient (83 WT) or -deficient (83 KO) intracranial xenografts.Supplementary Figure 12.Distinctive cellular landscapes of EGFR/EGFRvIII-proficient and -deficient brain tumours.Single cell RNA sequencing (scRNAseq) analysis of cell populations in intracranial xenografts driven by EGFR-expressing (GSC83-WT) and EGFRdisrupted (GSC83-KO) mesenchymal glioma stem cells.a. Distinct clusters of cells in EGFR-WT and EGFR-KO tumours; b.Differentially expressed genes (DEGs) in indicated tumours and cellular clusters; c.Gene expression levels of top-ranking marker genes in different murine populations identified in GSC EGFR-WT and EGFR-KO tumours; d.Transcriptional signatures of vascular and stromal cells in brain tumours; e.-f.tSNE plot of murine cells populations and UMAP plot of human glioma stem cells transcriptomes color coded by EGFR status.
. Gene expression profiles of stromal/murine cells in brain xenografts, as a function of EGFR/EGFRvIII status of cancer cells.a. Subsets of murine endothelial cells expressing human EGFR as determined by Cd31 and Cd34 co-expression (GSC83-EGFR-KO cells express dysfunctional but detectable EGFR transcript).b t-SNE plots of human glioma cells transcriptomes identify CD34 and EGFR colocalization in patient samples; c.Dot plot showing differential expression patterns of marker genes in individual clusters of endothelial cells.d.Dot plot showing differential expression patterns of marker genes in individual clusters of pericyte cell; e. Validation of human EGFR+ mouse endothelial cells in GSC83 tumours by cell sorting (FACS) and EGFR phosphorylation by magnetic bead separation (MACS; n=2 independent experiments).Human EGFR protein expressing Cd31+ endothelial cells in brain tumour of tumour EGFR status on myeloid and oligodendrocytic stromal cells in intracranial glioma xenografts in mice.a. tSNE plot of murine cells shows clustering based on gene expression.Three different populations of microglia and myeloid are detected; b. tSNE plot reveals different populations oligodendrocyte and their precursors.c.Dot plot showing differential expression patterns of marker genes in individual clusters of microglia and myeloid cells; d.Dot plot showing differential expression patterns of marker genes in individual clusters of oligodendrocyte and precursor cells; e. Relative proportion of oligodendrocyte and their precursor cells populations in either EGFR-KO or EGFR-WT tumours; f.Relative proportion of oligodendrocyte and precursor cells subtypes in either EGFR-KO or EGFR-WT tumours.

Filter
Units -100KDa-(Millipore # UFC905008) to a final volume of 1 mL.Concentrated conditioned medium was passed through 0.22μm filters and then centrifuged at 110,000 x g for 70 min.The resulting EV pellet was re-suspended in filtered 1 x PBS or RIPA buffer and stored at -80 o C until further use.

DEGs in GSC83 cellular clusters Clustering of single cell transcriptomes as a function of EGFR status UMAP_2
. All procedures involving animals were performed in accordance with the guidelines of the Canadian Council of Animal Care (CCAC) and the Animal Utilization Protocols (AUP) approved by the Institutional Animal Care Committee (ACC) at MUHC RI and McGill University (Protocol #5200).When possible, bioluminescence data was obtained using Xenogen (IVIS 200) bioluminescence scanner, and this was carried out after administering D-Luciferin Firefly potassium salt (Caliper Life Science;