Background.  Glioblastoma (GB), particularly IDH-wild type, is the most aggressive brain malignancy with a dismal prognosis. Despite advances in molecular profiling, the complexity of its tumor microenvironment and spatial organization remains poorly understood. This study aimed to create a comprehensive single-cell and spatial atlas of GB to unravel its cellular heterogeneity, spatial architecture, and clinical relevance.

Methods.  We integrated single-cell RNA sequencing data from 26 datasets, encompassing over 1.1 million cells from 240 patients, to construct GBmap, a harmonized single-cell atlas. High-resolution spatial transcriptomics was employed to map the spatial organization of GB tissues. We developed the Tumor Structure Score (TSS) to quantify tumor organization and correlated it with patient outcomes.

Results.  We showcase the applications of GBmap for reference mapping, transfer learning, and biological discoveries. GBmap revealed extensive cellular heterogeneity, identifying rare populations such as tumor-associated neutrophils and homeostatic microglia. Spatial analysis uncovered 7 distinct tumor niches, with hypoxia-dependent niches strongly associated with poor prognosis. The TSS demonstrated that highly organized tumors, characterized by well-defined vasculature and hypoxic niches, correlated with worse survival outcomes.

Conclusions.  This study provides a comprehensive resource for understanding glioblastoma heterogeneity and spatial organization. GBmap and the TSS provide an integrative view of tumor architecture in GB, highlighting hypoxia-driven niches that may represent avenues for further investigation. Our resource can facilitate exploratory analyses and hypothesis generation to better understand disease progression.

The classification of cancer has evolved over millennia, and centuries of work has laid the groundwork for modern cancer classification, which continues to evolve with advances in our understanding of cancer biology in tandem with improvement in the technologies, tools and frameworks used to characterize them. This thesis builds on the historical legacy of cancer classification by integrating single-cell transcriptomic approaches to explore the molecular complexity and intratumoral heterogeneity (ITH) of cancer. By defining and analyzing diverse transcriptomic states, known as metaprograms, in three aggressive cancer types—glioblastoma (GB), triple-negative breast cancer (TNBC), and diffuse midline glioma (DMG)—this work offers a more refined and precise lens through which to understand tumor progression and develop personalized therapeutic strategies. Using high-resolution single-cell RNA sequencing (scRNA-seq), spatially-resolved transcriptomics (SRT), and patient-derived organoid models, we identify distinct metaprograms that shape tumor progression, resistance, and patient outcomes.

Starting with DMG, we use spatial transcriptomics to map tumor-specific phenotypes, uncovering a novel neural stem cell-like population that interacts with the tumor microenvironment. This phenotype, defined by key progenitor markers, demonstrates plasticity, likely contributing to DMG’s resistance to therapy. By studying nonmalignant cells in the DMG microenvironment, we propose that specific cell types support tumor growth and evolution, highlighting potential therapeutic interventions. We then apply scRNA-seq to GB, revealing the presence of multiple metaprograms, including those linked to stem-like properties, invasion, and immune evasion. These metaprograms provide insights into how GB cells adapt and evolve in response to their microenvironment, uncovering potential therapeutic targets for this highly resistant cancer. In TNBC, we develop a comprehensive TNBC-Map by integrating single cell-datasets from patient biopsies, identifying nine core malignant metaprograms. These metaprograms encompass biological processes such as immune modulation, epithelial-to-mesenchymal transition (EMT), and vasculogenic mimicry. By correlating these metaprograms with patient survival, we identify distinct patterns of molecular activity that could guide the development of more personalized and effective treatments for TNBC.

Across these studies, we assess the power of metaprogram analysis to dissect cancer heterogeneity, offering a deeper understanding of the functional states driving tumor progression. This knowledge enables the identification of patient-specific molecular signatures, paving the way for precision medicine approaches. This thesis lays the groundwork for metaprogram-based cancer diagnostics and provides a foundation for the future integration of multi-omic precision medicine strategies that target specific cancer cell states, ultimately improving patient outcomes.

Elucidating spatiotemporal changes in gene expression has been an essential goal in studies of health, development, and disease. In the emerging field of spatially resolved transcriptomics, gene expression profiles are acquired with the tissue architecture maintained, sometimes at cellular resolution. This has allowed for the development of spatial cell atlases, studies of cell-cell interactions, and in situ cell typing. In this review, we focus on padlock probe-based in situ sequencing, which is a targeted spatially resolved transcriptomic method. We summarize recent methodological and computational tool developments and discuss key applications. We also discuss compatibility with other methods and integration with multiomic platforms for future applications.

Histone 3 lysine27-to-methionine (H3-K27M) mutations most frequently occur in diffuse midline gliomas (DMGs) of the childhood pons but are also increasingly recognized in adults. Their potential heterogeneity at different ages and midline locations is vastly understudied. Here, through dissecting the single-cell transcriptomic, epigenomic and spatial architectures of a comprehensive cohort of patient H3-K27M DMGs, we delineate how age and anatomical location shape glioma cell-intrinsic and -extrinsic features in light of the shared driver mutation. We show that stem-like oligodendroglial precursor-like cells, present across all clinico-anatomical groups, display varying levels of maturation dependent on location. We reveal a previously underappreciated relationship between mesenchymal cancer cell states and age, linked to age-dependent differences in the immune microenvironment. Further, we resolve the spatial organization of H3-K27M DMG cell populations and identify a mitotic oligodendroglial-lineage niche. Collectively, our study provides a powerful framework for rational modeling and therapeutic interventions.

Glioblastoma (GB), the most aggressive primary brain tumor, poses significant challenges due to its heterogeneity and resistance to current treatment modalities. This study investigates GB dynamics through a patient-derived organoid model to map cellular behaviors, including infiltration, migration, and transcriptional states. From a large cohort of patient-derived scRNAseq data, we identified 30 metaprograms encompassing key biological processes across GB samples using consensus Non-negative Matrix Factorization (cNMF). These metaprograms revealed distinct expression patterns and potential regulatory overlaps within malignant subpopulations, emphasizing their roles in GB development. By integrating temporal assessments, we traced metaprogram shifts in  our model, highlighting the plasticity of GB cells within the organoid environment. Comparisons with cortical and fetal brain studies and cancer hallmark signatures contextualized the findings, showing shared developmental and stress response features. This comprehensive approach underscores the potential of GB organoids for studying tumor behavior and improving the translation of therapeutic strategies.

Glioblastoma, isocitrate dehydrogenase (IDH)-wildtype (hereafter, GB), is an aggressive brain malignancy associated with a dismal prognosis and poor quality of life. Single-cell RNA sequencing has aided in grasping the complexity of the cell states and dynamic changes in GB. Large-scale data integration can help to uncover unexplored tumor pathobiology. Here, we resolved the composition of the tumor milieu and created a cellular map of GB (‘GBmap’), a curated resource that harmonizes 26 datasets, gathering 240 patients and spanning over 1.1 million cells. We showcase the applications of our resource for reference mapping, transfer learning, and biological discoveries. Reconstructing the tumor architecture using spatially resolved transcriptomics unveiled consistent niches across patients and their organizational gradient. The GBmap represents a framework that allows the streamlined integration and interpretation of new data and provides a platform for exploratory analysis, hypothesis generation, and testing. 

Diffuse Midline Glioma (DMG), particularly the H3-K27-altered, is a highly aggressivetumor with limited therapeutic options. Through spatial transcriptomics, we identified a novel NSC-like malignant phenotype that is characterized by the upregulation of key stem andprogenitor cell markers, including SOX2, OLIG1/OLIG2, TCF12, EPN2, MFGE8,FAM181B, PHLDA1, and PDGFRA. The surrounding tumor microenvironment (TME) resembled neurogenic niches, suggesting interactions fostering tumor plasticity.

Triple-negative breast cancer (TNBC), an aggressive and highly heterogeneous subtype of breast cancer, poses significant challenges for treatment due to its molecular diversity and resistance to standard therapies. Accounting for 10-20% of all breast cancer cases, TNBC lacks specific biological markers, making it difficult to classify and treat effectively. Traditional approaches based on bulk RNA sequencing obscure intratumoral heterogeneity and fail to capture distinct cellular states within tumors. In this study, we constructed a comprehensive single-cell transcriptomic map of TNBC by analyzing a cohort of published TNBC patient datasets, identifying nine transcriptomic states, or metaprograms, which capture the core behaviors of TNBC cells, including cancer stem cell properties, epithelial-to-mesenchymal transition (EMT), immune modulation, metabolic adaptation, and vasculogenic mimicry. We observed that these metaprograms are variably expressed within and across patient tumors, underscoring the complexity of TNBC. By integrating TNBC-specific metaprograms with established breast cancer subtypes, we found significant prognostic associations, with specific metaprograms correlating with poor survival outcomes. This study highlights the need for single-cell approaches to uncover TNBC’s molecular heterogeneity and suggests that metaprogram-based classification could facilitate more precise therapeutic interventions. Our findings provide a foundation for developing multi-omic strategies aimed at improving TNBC patient outcomes through personalized medicine.