false
Catalog
AANS Online Scientific Sessions: Tumor
Modeling the Genetic, Transcriptomic, and Cell Sur ...
Modeling the Genetic, Transcriptomic, and Cell Surface Antigen Heterogeneity of Glioblastoma Using Patient Derived Tumor Organoids
Back to course
[Please upgrade your browser to play this video content]
Video Transcription
Hello everyone, I hope you are doing well and staying safe. Thank you to the organizers for this special opportunity to share some of my research on glioblastoma organoids with you. Glioblastoma is an aggressive and deadly brain tumor, and it is famously known for its extensive heterogeneity. This heterogeneity exists between individual patients. For example, as illustrated by this figure, the somatic mutational landscape can vary from patient to patient. This heterogeneity also exists within one individual patient's tumor. Across relatively large anatomical distances, different tumor regions can have unique genetic and gene expression features. Here I highlighted a study that sampled one tumor at five different sites, which revealed five related but distinct molecular compositions. Finally, even at the microscopic level, tumor cells adjacent to one another can have different clonal origins and transcriptional states, which, as shown in this image, has been classified in a seminal paper by Neftel and co-workers into OPC-like, MPC-like, astrocyte-like, and mesenchymal-like states. This is a central problem in glioblastoma, where heterogeneity between and within patients at the genetic and transcriptomic levels lead to the persistent treatment resistance of this tumor despite all therapeutic attempts. We wanted to tackle this problem by developing a laboratory model of glioblastoma that better captured this heterogeneity. Our solution was to develop an organoid model of glioblastoma, where fresh tumor tissue is mechanically micro-dissected into 1 mm pieces and placed into suspension culture on an orbital shaker. Our simplified media omits EGF and FGF, which can preferentially favor cells with certain genetic backgrounds at the expense of others, leading to a loss of heterogeneity over time. These glioblastoma organoids, or GBOs, can form within 1-2 weeks, be expanded and passaged by mechanical cutting, and biobanked for long-term archiving and recovery at a later time. This method has a high success rate of more than 95% for primary glioblastomas and is broadly applicable to tumors of different genetic backgrounds. At the genetic level, we performed whole exome sequencing to demonstrate that GBOs maintain the somatic mutational landscape of their parent tumors. In this figure, between the GBO and tumor pairs, these mutations are present at similar allele frequencies, suggesting that the relative proportions of different genetic clones are largely maintained without selective expansion or depletion. We observed similar findings when we examined the structural variants where chromosomal regions were amplified or deleted to similar degrees in the GBOs compared with their parent tumors, as shown in these example genome tracks. From bulk RNA sequencing, we found that GBOs maintained high whole transcriptome-wide correlations with their parent tumors. Importantly, GBOs preserve patient-specific gene expression signatures and thus can model this aspect of intertumoral heterogeneity. To give some specific and potentially clinically relevant examples, I've highlighted GBOs from the anterior region of patient 7788's tumor, which contains high RNA and protein levels of EGFR, in contrast with GBOs from patient 7790's tumor, which maintains very low levels of EGFR. We were able to obtain samples from different regions of patient 7788's tumor and derive GBOs from the tissue. Interestingly, the posterior medial superior, or PMS, region of patient 7788's tumor contained high levels of EGFRv3, which is maintained in the GBOs, whereas the anterior region had almost no detectable levels of EGFRv3, which was also maintained in the GBOs. This intratumoral regional differences across relatively large anatomic distances is an important feature of glioblastoma. Patient 7788's tumor was sampled in five different regions, anterior, medial, lateral, posterior, and posterior medial superior, and patient 7884's tumor was sampled in two different areas, frontal and temporal. We took an unbiased and more global approach to analyze the GBOs derived from these different anatomic regions, and we found that the GBOs largely maintained the regional gene expression signatures. At the microscopic level, GBOs maintained the cellular composition of their parent tumor. H&E staining demonstrated that the GBOs maintained the global cellularity and stromal composition of their parent tumors. In addition, GBOs from patient 7966 contained gemistocytic cells, whereas GBOs from patient 8036 contained large multinucleated cells, both of which were unique to their parent tumors. We also performed immunofluorescence for cell-type specific markers, demonstrating that the populations of different types of tumor cells, such as SOX2-expressing or Olig2-expressing cells, were maintained at similar abundances over time in culture. Notably, cell proliferation, as marked by Ki-67 immunoreactivity, was maintained at a modest 10-30%, reflecting the native proliferation rate of the tumor. This is in contrast to many cell culture models where rapidly dividing cells are enriched for over time, causing depletion of slowly cycling or senescent cells. These slowly cycling or senescent cells may play an important role in the tumor cell composition and microenvironment, with a few potentially serving as quiescent stem-like cells that reactivate after treatment, leading to tumor recurrence. Examining these cells beyond their canonical markers, we also performed single-cell RNA sequencing and classified them according to the novel four-state model proposed by Neftal and coworkers. As examples, this figure shows that the tumor and GBOs from patient 8036 contain all four types of cells, whereas the tumor and GBOs from the core region of patient 8165's tumor consisted of largely astrocyte-like and mesenchymal-like cell states. Also at the single-cell level, we examined the cell surface expression of EGFRV3, an important tumor oncogene and antigen for immunotherapies. And GBOs derived from tumors that were all clinically determined to be EGFRV3-positive. Using non-neoplastic cells as our negative control, we found quite a bit of variability in the percentage of cells that expressed EGFRV3, ranging from 5 to 90% of the tumor cells. And this highlights, perhaps, one of the major challenges facing EGFRV3-targeted therapies. Likely the most unique feature of our GBO system is the preservation of native non-neoplastic cell types that make up the tumor microenvironment. Single-cell RNA sequencing of tumors and corresponding GBOs from three different patients revealed populations of stromal cells, including endothelial cells and pericytes, oligodendrocytes, microglia, macrophages, and T-cells. As an example, this immunofluorescence image highlights the microglia and macrophages, as marked by IBO1, as well as T-cells, as marked by CD3, in one pair of tumor and GBOs. Focusing on the macrophages, which are prevalent and an essential part of the tumor microenvironment, we found that the expression of essential marker genes and cytokines in the GBOs were largely maintained, suggesting that these macrophages might actually be functionally preserved in the GBOs. GBOs maintain the native heterogeneity of glioblastoma in a laboratory tumor model that is readily amenable to many avenues of experimentation. We have been able to use GBOs to perform lineage tracing experiments to understand mechanisms of tumor genesis, screen novel small-molecule targeted therapies based on the tumor's genetic background, test emerging immunotherapies such as CAR-T cell therapy, and study tumor cell migration after orthotopic xenograft into immunodeficient mice in search of cells and pathways responsible for this invasive behavior. With the fidelity and flexibility of this tumor model, I hope that GBOs can find its place in the neuro-oncology research toolbox to help advance the field towards a much-needed cure for this deadly disease. I'd like to end by acknowledging the incredible mentorship from Professor Hongjun Song and Dr. Donald O'Rourke, support from a fantastic team within the lab and our numerous clinical collaborators at Penn, and of course, the essential support from my funding sources. Thank you for listening to my presentation and for taking part in the virtual version of the AANS 2020 Annual Meeting Tumor Session.
Video Summary
The video transcript summarizes research on glioblastoma organoids, which are laboratory models of glioblastoma, an aggressive and deadly brain tumor. Glioblastoma is known for its heterogeneity within and between patients, both at the genetic and transcriptomic levels, leading to treatment resistance. The researchers developed a method to create glioblastoma organoids that capture this heterogeneity by culturing fresh tumor tissue. These organoids maintain the genetic and transcriptional landscapes of their parent tumors and preserve patient-specific gene expression signatures. They also maintain the cellular composition and microenvironment of the tumor, making them a valuable tool for studying tumor genesis, testing therapies, and advancing neuro-oncology research. The speaker acknowledges their mentors, collaborators, and funding sources. (194 words)
Asset Subtitle
Daniel Zhang
Keywords
glioblastoma organoids
laboratory models
brain tumor
heterogeneity
treatment resistance
×
Please select your language
1
English