Human organotypic brain slice culture: a novel framework for environmental research in neuro-oncology

(A) The relative separation of the tumor-infiltrated cortex from the healthy cortex was verified by means of hematoxylin and eosin staining. Microvascular proliferation (black arrows) is seen in (i) tumor tissue compared with (ii) healthy cortex. (B) Illustration of the workflow for the generation of the sections used for organotypic brain cultures. A vibratome was used to obtain 18–20 coronal sections of 300 μm thickness from every 1 × 2-cm tissue block, with dimensions of ∼10 × 15 mm. Sections were then transferred to the nylon membrane of an insert in a six-well plate using the blunt, fire polished end of a glass Pasteur pipette and incubated at 37°C with the surface of the media contacting the membrane, enabling diffusion for up to 14 d. Sections were then characterized at different time points using (i) MEA electrophysiology to confirm neuronal activity, (ii) immunohistochemistry to evaluate the cell composition and loss over time, (iii) ELISA to evaluate the cellular damage or cytokine measurement, and (iv) MinION RNA-sequencing/qPCR to evaluate profile changes in gene expression.

Acute sections were collected from each batch as control. Sections cultivated for weeks were then compared with acute sections, with N = 20 patients. (A) From left to right: 5× tiled representative fluorescence images of tissue sections for Acute, DIV1, DIV4, DIV7, and DIV14. Sections were costained against the neuronal marker (NeuN), astrocyte marker (GFAP), and nuclei (DAPI Fluoromount). Scale bars are 1 mm; representative 5× image of each time point showing the cytoarchitecture of the slice. Scale bars are 200 μm; representative 20× image of each time point showing neurons (NeuN), astrocytes (GFAP), and nucleus (DAPI). Scale bars are 50 μm. (B–E) Quantification of NeuN (neurons) positive staining demonstrated a significant difference in the distribution of neuronal sections between fresh versus cultured sections (P < 0.0001). (C) Quantification of rate of neuronal loss showed a negligible reduction over DIV1–14. (D) Quantification of GFAP (astrocytes) shows an increase on fresh sections versus DIV1 because of injury (P < 0.00001), and (E) quantification of astrocytes loss shows a nonsignificant difference from DIV1 to 14 (DIV4 = −5.97%, DIV7 = −8.15%, and DIV14 = 5.58%, one-way ANOVA with Bonferroni correction for multiple comparisons). (F) Cellular necrosis in the brain slice cultures was done using a TUNEL assay for N = 3 patients at different time points. (G) Quantification of TUNEL-positive cells shows a nonsignificant difference from DIV1 to 14 (DIV4 = 6.74%, DIV7 = 11.35%, and DIV14 = 0.613%). (H) Cellular metabolism was quantified using an LDH assay in different time points. (I) Quantification of LDH assay shows a nonsignificant difference from DIV1 to 14 (DIV4 = 33.87%, DIV7 = 34.51%, and DIV14 = 26.31%). Statistics were performed using unpaired t test with Welch’s correction. *P < 0.05, **P < 0.001, ***P < 0.0001, ****P < 0.00001, ns = nonsignificant. A representative result of three independent experiments is shown (error bar represents ± SD).

(A) Illustration of the experimental workflow. Sections were cultured for up to DIV14, and recordings were performed to assess electrophysiological activity. The sections were placed on the recording array as illustrated. Neuronal activity was evoked by means of perfusion with high K⁺ medium. (B) Recorded data were high-pass filtered (300 Hz), and neuronal events were extracted and averaged before plotting (red: DIV1, blue: DIV14). (C) Random sampling of the extracellular electrophysiological recording.

(A) Workflow of an RNA-seq experiment with ∼500 mg tissue using Nanopore MinION. After the culture period specified, the tissue was homogenized, RNA extracted, amplified, the library prepared, and sequenced from N = 3 patients. (B) Differential bulk gene expression showing top 500 up- and down-regulated genes. The top 500 up- and down-regulated genes lay within the same expression region. (C) Comparison of the gene expression profile at different time points. The gene expression profile exhibits a change in expression from the acute to DIV7. However, the expression profile remains stable when compared between DIV7 and DIV14. (D–G) Cell-specific expression profiles of neurons, oligodendrocytes, astrocytes, and microglia. Key genes from each cell type are shown in the dotted box next to the expression plot. The cell-specific signatures were chosen based on their fidelity score. The scale bar for y-axis is global as represented in (B).

(A) Illustration of the experimental workflow showing tumor injection on DIV1, and tumor progression was further validated using imaging. (B) (i) proliferation pattern of the TMZ-sensitive proneural GSC_CL1 over 14 d of culture. (ii) The proliferation pattern is disrupted when TMZ (50 μM) is added to the culture medium. TMZ was added to the medium 3 d postinjection of GBM cells into the slice (DIV4). (C) Scale bars are 200 μm. (C) (i) proliferation pattern of the mesenchymal cell type GSC_CL2 over 14 d. (ii) The proliferation is uninterrupted when TMZ (50 μM) is added to the culture medium. TMZ was added to the medium 3 d postinjection of GBM cells into the slice (DIV4). Scale bars are 200 μm. (D) Quantification of the increase in the area of the injected tumor for GSC:CL1 and GSC:CL2 shows +59.2% fold change for GSC:CL2 while +32.3% for GSC:CL1, P < 0.0001, and in the presence of 50 μM TMZ, GSC_CL1 shows halted growth, whereas the GSC_CL2 shows a minor response to the treatment. (E) 3D imaging and reconstruction was performed using two-photon microscopy. The left images demonstrate the tumor core in GSC_CL1 and GSC_CL2, and the right images represent the infiltrative margin of the tumor for GSC_CL1 and GSC-CL2. Scale bars are 50 μm. Unpaired t test with Welch’s correction were used.

The cytokine profiling shows that there are significant changes in sections because of the development of the TME using one-way ANOVA test with Bonferroni’s multiple comparison test. *P < 0.05, **P < 0.001, ***P < 0.0001, ****P < 0.00001.

(A) Astrocyte extraction protocol from the cultured sections. MACS was used to extract the astrocytic cells with biotin-labelled antibody. (B) qPCR-based heat map of signature astrocytic genes that mark the different reactive state of the extracted astrocytes. Signature genes were extracted from publicly available datasets. (C) Expression analysis of reactive marker genes selected by specificity for humans and astrocytic fidelity score, in astrocytes purified from GBM patients and our human slice model with tumor injection.