Cell type transcriptional identities are maintained in cultured ex vivo human brain tissue

Abstract

It is becoming more broadly accepted that human-based models are needed to better understand the complexities of the human nervous system and its diseases. The recently developed human brain organotypic culture model is one highly promising model that requires the involvement of neurosurgeons and neurosurgical patients. Studies have investigated the electrophysiological properties of neurons in such ex vivo human tissues, but the maintenance of other cell types within explanted brain remains largely unknown. Here, using single-nucleus RNA sequencing, we systematically evaluate the transcriptional identities of the various cell types found in six patient samples after fourteen days in culture (83,501 nuclei from day 0 samples and 45,738 nuclei from day 14 samples). We used two pediatric temporal lobectomy samples, an adult frontal cortex sample, two IDH wild-type glioblastoma samples, and one medulloblastoma sample. We found remarkably high correlations of day 14 transcriptional identities to day 0 tissue, especially in tumor cells (r = 0.90 to 0.93), though microglia (r = 0.86), oligodendrocytes (r = 0.80), pericytes (r = 0.77), endothelial cells (r = 0.78), and fibroblasts (r = 0.76) showed strong preservation of their transcriptional profiles as well. Astrocytes and excitatory neurons showed more moderate preservation (r = 0.66 and 0.47, respectively). Because the main difficulty with organotypic brain cultures is the acquisition of human tissue, which is readily available to neurosurgeons, this model is easily accessible to neurosurgeon-scientists and neurosurgeons affiliated with research laboratories. Broad uptake of this more representative model should prompt advances in our understanding of many uniquely human diseases, lead to more reliable clinical trial performance, and ultimately yield better therapies for our patients.

Keywords: Human brain; brain model development; epilepsy; glioblastoma; human brain organotypic slice culture; medulloblastoma.

Figure 1.. The human brain organotypic slice culture model is feasible and shows conservation of most cell types across time.

A, Workflow of the human brain organotypic slice culture model, from MRI to tissue sectioning to culture plates to single-nucleus RNA sequencing. B, List of tissues used in this study with anatomic locations and pathologies, as well as approximate ages. C, Bar graphs showing the relative proportions of identified cell types in day 0 and day 14 samples for each patient tissue. Cell types were identified using canonical marker gene expression from published single-cell and single-nucleus RNA sequencing data. Number of nuclei that passed quality control and were used for analysis listed for each time point.

Figure 2.. Single-nucleus RNA sequencing shows retained cell type transcriptional identities in tissues with more normal cell types.

A, Integrated UMAP projections of Day 0 and Day 14 cells for tissues A, B, and C. Each point represents a single cell. B, Split UMAPs color-coded by cell type for day 0 and day 14 cells for tissues A, B, and C. Day 0 and day 14 cell types generally cluster together though there are some differences, e.g. tissue C’s neurons in day 0 and day 14. Cell types assigned using published marker genes (see methods). C, Correlation matrices comparing the log2 fold change of a given cell type at day 14 to cell types at day 0. Log2 fold changes generated by comparing a given cell type at day 0 to the rest of the day 0 cells, and given cell types at day 14 to the rest of the day 14 cells. Higher correlations at intersections of cell types indicate that the unique transcriptional profile differentiating that cell type from other cell types at day 0 is better preserved at day 14.

Figure 3.. Cells within tumor tissues retain their unique transcriptional profile after two weeks in culture.

A, Split UMAPs for glioblastoma tissues D and E color-coded by cell types. Some cells were not able to be assigned clearly to a cell type and so left unlabeled (Tissue D, bottom left cluster in lighter grey). B, Split UMAP for medulloblastoma tissue (tissue F) color-coded by cell types. C, Correlation matrices comparing day 0 to day 14 cell types for tissues D (IDH wild-type glioblastoma), E (IDH wild-type glioblastoma), and F (SHH-type medulloblastoma). D, Graph showing the individual values for each tissue sample and the overall average correlation coefficient between day 0 and day 14 cell types. Error bars are standard error of the mean (n = 1–6).