Fig 1.
Dissecting the transcriptional heterogeneity of tumor-associated astrocytes in experimental mouse gliomas.
(A) UMAP showing brain and tumor cell projection by spot clustering, spatial mapping on the tissue section with the Visium 10× Genomics platform and matching with H&E staining from the same tissue section. (B) Identification and segregation of distinct areas being readouts from UMAP on spatial projection as the tumor core (TC), tumor periphery (TP), border (BO), and brain regions (BR). (C) Spatial mapping of astrocyte marker genes (Gfap, Aldh1l1, and S100b) on tissue sections from Visium gene expression readout. (D) Violin plots showing expression profiles of Gfap, Aldh1l1, and S100b at different tumor and nontumor regions of the tumor microenvironment. n = 2 per condition. (E) Spatial mapping of Rab6a (pan astrocytic marker), Aqp4 (water channel marker), Pdpn (astrogliosis marker), Lcn2 and Sipr3 (neuroinflammatory markers).
Fig 2.
Distinct morphology of astrocytes in various tumor areas.
(A) Immunofluorescence (IF) staining of mice brain sections showing td tomato labeled tumor cells (in red), Gfap+ astrocytes (in green) and DAPI stained cell nuclei (in blue) depicts a “glial scar’” composed of reactive astrocytes surrounding the tumor core. Gfap+ astrocytes are barely present in the TC (in the magnified inset), show elongated shapes at the BO and star-like morphology in the surrounding brain parenchyma, BR. (B) IF staining of S100b+ astrocytes (green) shows different patterns, with a uniform, widespread distribution in the brain and the presence of S100b+ astrocytes in the TC and at the BO. S100b+ astrocytes in the BO are elongated, while S100b+ astrocytes farther from the TC display star-like morphology. (C) The scheme illustrates a distinct morphology of astrocytes in BO and BR areas. (D) Double IF staining shows S100b+ and Gfap+ bipolar-extended astrocytes at the BO, whereas in the BR astrocytes display radial-star shapes. (E) IF staining shows Aldh1l1+, bipolar-extended astrocytes (in green) in the TC and BO regions (indicated by arrows); n = 4 per condition.
Fig 3.
Loss of neurons and glutamate dysregulation coincides with A1 and A2 astrocyte signatures at the tumor core.
(A) NeuN+ neurons (in green) are absent in the tumor core demarcated by the presence of td tomato tumor cells (in red), while NeuN+ neurons are distributed uniformly outside of the TC. (B) Spatial mapping of neuronal markers such as Rbfox3, NeuN, Gria1, and glutamate transporter 1 marker Slcia2/Glt1 in tissue sections confirms a lack of neuronal gene expression and Slcia2 at the TC. (C) IF staining of GLT-1 (in green) and astrocyte marker S100B (blue) in mouse brain section with td tomato tumor cells (in red). The arrows indicate numerous double-positive GLT-1+ and S100B+ astrocytes at the tumor periphery (TP) whereas no double-positive cells are visible within the TC; separate channels representing the indicated co-staining. This shows dysregulation of GLT-1 at the TC but not at TP; n = 4 per condition. (D) Spatial correlation extracted from the Visium data indicates stronger correlation between astrocyte genes such as Gfap, Aldh1l1, and S100B with Slc1a2 at the TP, BO, and BR compared to the TC; n = 2 per condition. (E) Spatial mapping of A1 astrocyte-specific genes (Cfb, C3) and A2 astrocyte-specific genes (S100a10, Sphk1) on tissue sections from the Visium gene expression readout. (F) Spatial correlation from the Visium gene expression data indicates correlation of cells expressing C3 with astrocytic markers at different tumor regions. The underlying data is available at S1 Data.
Fig 4.
Proximity of microglia influences TAAs in GBM.
(A) IF staining reveals TMEM 119+ microglia (in green) with a radial shape (surveying microglia) exclusively in BO and BR regions, whereas microglia with ameboid shapes (activated microglia) are present at the TC and mostly within TP regions, close to td tomato tumor cells (in red). In the insets; TMEM119+ microglia in BO and TC regions show morphological diversity; n = 4 per condition. (B) Spatial mapping of TMEM119-expressing microglia in and around the TC, whereas microglia expressing Sall1 (which is exclusively expressed in surveying microglia) are distributed outside the TC; n = 2 per condition. (C) Spatial mapping of Tnfα, Il1β, and C1qa depicts localization in the TC. (D) Scheme showing how tumor cells and TAAs influenced by microglia can orchestrate the elimination of neurons at the TC. (E) Schematic representation of 5 subtypes of TAAs localized within the tumor-bearing brain based on marker gene expression depicted by spatial transcriptomics and to some extent validated by immunofluorescence.
Fig 5.
Spatial transcriptomic analysis of human glioblastoma samples confirms distinct tumor microenvironment regions.
This figure presents a comprehensive spatial transcriptomic analysis of human glioblastoma samples using the Banksy computational framework. (A) UMAP visualization of integrated transcriptomic data from six (n = 6) GBM samples, colored by Leiden clustering at resolution 0.75. Distinct clusters are numbered 1–17. (B) Spatial distribution of identified clusters across six GBM samples (GBM1-GBM5_2), showing the topographical organization of transcriptionally distinct regions within each tumor section. (C) Heatmap comparing GBM clusters with mouse GL261 glioma regions, showing the relative similarity (scaled cosine similarity) of human GBM clusters to predefined mouse tumor regions classified as “brain,” “border,” or “tumor,” Clusters 3, 6, and 17 show the highest similarity to normal brain tissue; clusters 1, 5, 10, and 11 correspond to tumor–brain border regions; other clusters primarily match the tumor core n = 6. (D) Dot plot showing expression of marker genes across identified clusters, organized by functional groups and by spatial region. Dot size represents percentage of cells expressing each gene, while color intensity indicates average expression level. Notable patterns include A1 reactive astrocyte markers in border regions and tumor-specific expression of ECM-related genes, n = 6.
Fig 6.
Mapping extracellular matrix remodeling in experimental gliomas.
(A) Spatial mapping of Tenascin C, Fibronectin 1, Laminin b1, Integrin alpha1 encoding important extracellular matrix proteins in sections of brains bearing gliomas denotes localization of these markers in and around TC and TP regions. (B) IF staining shows upregulation of Fibronectin and Laminin in the TC, in agreement with spatial transcriptomics data. (C) Upregulation of Tgm2 encoding a tissue transglutaminase 2 in sections of brains bearing gliomas whereas there is no expression of Tgm2 in brains of naïve mice. Spatial mapping shows that expression of Tgm1 is restricted to the BO region of the tumor. (D) Schematic representation of extracellular matrix alterations in GBM undergoing from a soft to stiff tissue could be regulated by upregulation of ECM proteins such as Tenascin C, Fibronectin, Integrin along with upregulation of a crosslinking protein such Tgm2; n = 2 per condition (spatial transcriptomics), n = 4 per condition (immunofluorescence).
Fig 7.
TAAs up-regulate Tgm2 in glioma cells.
(A) Primary cultures of murine astrocytes and GL261 glioma cells were grown separately or together for 24 h followed by immunohistochemistry and western blot analysis, n = 4. (B) Representative IF images of astrocytes and GL261 cells showing Tgm2 staining. (C) Western blot analysis for Tgm2 from cell lysates of astrocytes, GL261 cells and from those cells growing in co-culture. (D) Quantification of Tgm2 protein levels by densitometry of blots demonstrates increased Tgm2 levels in GL261 cells induced by co-culture with astrocytes. Tgm2 levels were analyzed by western blot and densitometry of immunoblots determined from 4 experiments is represented as mean± SD. (E) Co-culture of astrocytes with NRas of PDGF glioma cells increases laminin levels in astrocytes. (F) Densitometric analysis of immunoblots from three experiments. Data were normalized to the levels of GAPDH in the same sample; control is set as 1; P values were calculated using GraphPad on logarithmic values and considered significant when *P < 0.05 (one-way paired t test). The underlying data is available at S2 Data.
Fig 8.
Pharmacologic modification of the tumor microenvironment reduces astrogliosis and harmonizes astrocytic GLT1 expression.
(A) In RGD-treated mice less Gfap+, reactive astrocytes are detected in the astrocytic ring around the TC and more Gfap+, reactive astrocytes are present in the TC compared to controls. (B) Quantification of percentage of area occupied by Gfap+, reactive astrocytes/DAPI at TC, BO, and BR regions. Statistical significance was calculated with one-way paired t test, *p < 0.05; **p < 0.01, n = 5 per group/condition. The underlying data is available at S3 Data. (C) Schematic representation of astrocytic heterogeneity represented by Gfap+, reactive astrocytes at TC, BO, and BR regions and modulation of their distribution in RGD-treated mice. (D) Representative image of GLT-1+ with S100b+ TAAs in brain sections of RGD-treated mice depicts harmonization of GLT-1 expression in S100b+ TAAs; n = 4 per group/condition. (E) Scheme showing that the blockade of integrin signaling by the RGD peptide impacts microglia functions and disrupts their crosstalk with TAAs, which prevents misusing of those cells in tumor progression.