Fig. 3: Comparative transcriptome analysis of GIC/iNSC identifies a GAG-mediated mechanism of Tregs migration in a proportion of GBM.

a Network representation of the enriched pathways identified in the 10 syngeneic GIC-iNSC comparisons. Nodes represent pathways; pie charts are shaded according to the patients in which the pathway is identified (see legend); size is proportional to the mean value of −log10(FDR) across all patients. Edges represent deregulated genes shared between pathways (pooled over relevant patients); thickness is proportional to the number of genes shared. b Heatmap showing the Spearman rank correlation across patients between pathway enrichment in the GIC culture relative to matched iNSC, expressed as –log10(p), and estimated cell type composition in the corresponding GBM bulk tissue sample. Cells with statistically significant correlation coefficient are highlighted (p < 0.05). c Representative histology of glioblastoma bulk tissue (Hematoxylin & Eosin (H&E) staining, top), immunostaining for CD4 (bottom left) and FoxP3 (bottom right). FoxP3/CD4 ratio in all GBM samples (right). Each case has been stained once. Scale bar is 250 µm. d Disaccharide composition of CS isolated from GIC (black bars) and iNSC (grey bars) cell extracts (plain bars) and from conditioned media (hatched bars) analyzed by RP-HPLC. D0a0: HexA-GalNAc; D0a6: HexA-GalNAc(6S); D2a0 HexA(2S)-GalNAc; D0a4: HexA-GalNAc(4S); D2a6 HexA(2S)-GalNAc(6S); D0a10: HexA-GalNAc(4S, 6S). Results are an average of two GIC/iNSC pairs, 19 and 31, (n = 2, two way ANOVA). (D0a6: p value (GIC vs iNSC cells/matrix) < 0,0001; p value (GIC vs iNSC CM) = 0,0009; D2a0/D0a4: p value (GIC vs iNSC cells/matrix) <0,0001, p value (GIC vs iNSC CM) = 0,0003). Percentages of migrated immune cells in a transwell assay in the presence of GIC (black bars), their 24 h conditioned-media (CM, hatched black bars), iNSC (grey bars) and media only (light grey bars) after 4 h incubation. Percentage of Tregs (CD4+CD8-FOXP3+CD25+CD127−, e and other immune cells (CD14+ monocytes, CD56+natural killer cells, CD4+Foxp3−Cd25− and CD8+Foxp3− cells, f with (orange border) and without chondroitinase ABC (chABC) pre-treatment. Results are obtained from patients 19 and 31, each experiment was repeated 4−7 times (GIC media: n = 6, NSC media: n = 4, GIC: n = 7, GIC CM: n = 5, GIC + Chase: n = 7, GIC CM + Chase: n = 5, iNSC: n = 5, one-way ANOVA). e p value (Media vs GIC cells) < 0,0001; p value (Media vs GIC CM) = 0,0103; p value (GIC cells vs chABC) < 0,0001; p value (GIC CM vs chABC CM) = 0,0184; p value (GIC cells vs GIC CM) = 0,0806. f CD4⁺Foxp3⁻CD25⁺: p value (Media vs GIC) = 0,0004, p value (Media vs GIC chABC) = 0,0002. g Schematic of experimental workflow (top left), percentage of GFP-labelled hTregs over total cell number per tumour section upon treatment with vehicle (control) or chondroitinase ABC (chABC) (bottom left), (n = 4, two tailed t-test p value = 0,0049). Representative GFP immunofluorescence, scale bar is 20 µm. All graphs report mean ± SEM. Statistical significance for all panels *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Source data are provided in the source data file.