Modeling glioblastoma invasion using human brain organoids and single-cell transcriptomics

Teresa G. Krieger, Stephan M. Tirier, Jeongbin Park, Tanja Eisemann, Heike Peterziel, Peter Angel, Roland Eils, Christian ConradbioRxiv 2019 · DOI

TL;DR

This paper introduces a novel in vitro model using human brain organoids to study glioblastoma (GBM) invasion, revealing how tumor cells interact with healthy brain tissue and adapt their gene expression. This model, combined with single-cell transcriptomics, identifies potential therapeutic targets by uncovering key gene expression changes and cell-cell communication pathways involved in GBM's aggressive spread.

What Problem Does This Paper Address?

Glioblastoma (GBM) is a highly aggressive brain cancer with a poor prognosis, largely due to its invasive nature. Understanding and treating this invasion is challenging because GBM cells are highly heterogeneous, and existing in vitro models often lack the complexity and physiological relevance of the human brain. Specifically, it's difficult to study how GBM cells interact with normal brain cells and how these interactions drive invasion and alter tumor cell behavior in a way that's relevant to clinical timelines. This paper aims to bridge this gap by creating a more realistic in vitro system to dissect GBM invasion and its underlying molecular mechanisms.

Methods

The researchers developed a sophisticated experimental system combining several cutting-edge techniques:

  • Human Brain Organoids: They used induced pluripotent stem cell (iPSC)-derived human cerebral organoids as a 3D scaffold. These organoids mimic the complex cellular architecture of the developing brain, providing a more physiologically relevant environment than traditional 2D cultures.
  • GBM Cell Invasion Assay: Patient-derived GBM cells were co-cultured with these organoids.
  • Tissue Clearing and Confocal Microscopy: To visualize the intricate invasion process, they employed tissue clearing techniques (specifically the FRUIT protocol) followed by confocal microscopy. This allowed them to image deep within the organoids and observe tumor cell behavior in 3D.
  • Semi-Automated Image Analysis: A custom workflow was developed to quantitatively analyze invasion depth, tumor cell spread, and the formation of tumor microtubes (long cellular extensions used for communication and invasion).
  • Single-Cell RNA Sequencing (scRNA-seq): This was a crucial component. They performed scRNA-seq on GBM cells before and after co-culture with organoid cells. This allowed them to pinpoint specific transcriptional changes that occur upon interaction with the brain microenvironment.
  • Ligand-Receptor Interaction Analysis: Using the scRNA-seq data, they computationally screened for potential molecular interactions (ligand-receptor pairs) between GBM cells and organoid cells, aiming to identify key communication pathways driving invasion.

Datasets

The study utilized:

  • Patient-Derived Glioblastoma Cell Lines: Four distinct GBM cell lines from different patients were used to assess intertumoral heterogeneity.
  • Human iPSC-Derived Cerebral Organoids: These served as the in vitro model of healthy brain tissue.
  • scRNA-seq Data: Transcriptomic data was generated from GBM cells cultured alone and from GBM cells co-cultured with organoid cells. This data was deposited in the European Genome-Phenome Archive (EGA).

Key Insights and Ideas

  • Recapitulating in vivo Invasion: The organoid model successfully replicated key aspects of GBM invasion observed in actual tumors, including the formation of long microtubes and invasion into specific brain tissue layers. This validates the model's physiological relevance.
  • Coherent Transcriptional Adaptation: Despite significant intertumoral heterogeneity (differences between patients), GBM cells showed a remarkably consistent set of gene expression changes when interacting with organoid cells. This suggests that GBM cells have a conserved response to the brain microenvironment, upregulating genes essential for invasion and network formation.
  • Identifying Key Genes for Invasion: The study pinpointed specific genes, such as GJA1 (connexin-43, involved in gap junctions) and GPC3 (a cell surface proteoglycan), that were consistently upregulated and are implicated in GBM's invasive behavior and network formation.
  • Uncovering Communication Pathways: The ligand-receptor analysis revealed potential molecular dialogues between GBM and organoid cells, highlighting specific interactions that could be targeted therapeutically. This provides a roadmap for developing more precise treatments.

Key Results

  • Extensive Microtube Formation: GBM cells in organoids extended microtubes up to 450 µm long, mirroring in vivo observations.
  • Invasion Depth: Tumor cells invaded up to 100 µm into the organoids, with some reaching approximately 300 µm from the surface.
  • Intertumoral Heterogeneity in Invasion: The model captured differences in invasive capacity between GBM cell lines from different patients.
  • 45 Consistently Upregulated Genes: A core set of 45 genes were upregulated in GBM cells across all patient samples upon co-culture with organoid cells, indicating a shared response to the brain environment. These genes were enriched for functions related to growth regulation, neuronal migration, and secretion.
  • Identification of Potential Therapeutic Targets: The ligand-receptor analysis identified several candidate pairs, including collagen-integrin interactions and pathways involving glypican-3 and Notch signaling, that are upregulated upon co-culture and could be targeted.

Limitations and Future Work

The authors acknowledge that while this model is powerful, it's still an in vitro system.

  • Limited Patient Samples: While four patient-derived cell lines were used, more samples would be beneficial to robustly link intertumoral differences in invasion to specific transcriptional changes.
  • Further Functional Validation: The identified ligand-receptor pairs are putative interactions. Future work should focus on experimentally validating their functional significance in GBM invasion.
  • Expanding to Other Tumors: The model's applicability to studying invasion in other types of brain tumors or even other cancers could be explored.
  • Live Imaging and Advanced Modalities: The authors suggest that the organoid model is amenable to live imaging (two-photon, light-sheet microscopy) and integration with other single-cell sequencing modalities (like chromatin accessibility) to gain a more comprehensive understanding of GBM's functional, transcriptional, and epigenetic factors driving invasion.
  • Drug Screening Platform: The model holds promise as a platform for high-content drug screens to assess patient-specific drug responses, potentially accelerating the identification of effective treatments.

Summary updated

Abstract

Abstract Glioblastoma multiforme (GBM) are devastating neoplasms with high invasive capacity. GBM has been difficult to study in vitro . Therapeutic progress is also limited by cellular heterogeneity within and between tumors. To address these challenges, we present an experimental model using human cerebral organoids as a scaffold for patient-derived glioblastoma cell invasion. By tissue clearing and confocal microscopy, we show that tumor cells within organoids extend a network of long microtubes, recapitulating the in vivo behavior of GBM. Single-cell RNA-seq of GBM cells before and after co-culture with organoid cells reveals transcriptional changes implicated in the invasion process that are coherent across patient samples, indicating that GBM cells reactively upregulate genes required for their dispersion. Functional therapeutic targets are identified by an in silico receptor-ligand pairing screen detecting potential interactions between GBM and organoid cells. Taken together, our model has proven useful for studying GBM invasion and transcriptional heterogeneity in vitro , with applications for both pharmacological screens and patient-specific treatment selection at a time scale amenable to clinical practice.
BibTeX
@misc{krieger2019modeling,
  author = {Teresa G. Krieger and Stephan M. Tirier and Jeongbin Park and Tanja Eisemann and Heike Peterziel and Peter Angel and Roland Eils and Christian Conrad},
  title = {{Modeling glioblastoma invasion using human brain organoids and single-cell transcriptomics}},
  year = {2019},
  doi = {10.1101/630202},
  howpublished = {bioRxiv},
}
Figure