Cancer Biology

Understanding tumor development, cancer progression, and immune response using high-resolution spatial transcriptomics

Studying tumor heterogeneity and the complexity of the tumor microenvironment has always been challenging with conventional methods of analysis and continues to be an active research field that adapts quickly to new technology. Pinpointing the locations of different cell populations within a tumor mass and identifying the gene expression signatures of the cells in a tumor neighborhood would greatly improve our understanding of tumor progression, metastasis, tumor immunology, and how the tumor microenvironment affects all of these processes. The Curio Seeker Spatial Mapping Kit enables researchers to map cellular neighborhoods in the tumor at single-cell scale and how they change during development, disease, treatment, and repair.  

The following papers highlight how analysis with Slide-seq, the foundational technology behind Curio Seeker, provides new insights into different research areas in cancer biology.

Tumor microenvironment

TUMOR MICROENVIRONMENT

Dissecting the human prostate tumor microenvironment

Hirz et al., Nature Communications, Jan 2023.

In this paper, Hirz et al. combined single-cell RNA-seq and spatial transcriptomics to study the human prostate tumor microenvironment. Specifically, the authors utilized a tissue dissociation protocol to enrich for immune cells to facilitate in-depth analysis with single-cell RNA-seq, and integrated the analysis with Slide-seq, which did not involve any tissue disruption and preserved tissue architecture, to provide a holistic picture of the tumor microenvironment. Here, we focus on their findings from Slide-seq. 

The authors applied Slide-seq to healthy prostate tissues, low-grade (LG) tumors, high-grade (HG) tumors, and adjacent non-tumor prostate tissues. Using the algorithm RCTD (Robust Cell Type Decomposition) with a scRNA-seq reference, each Slide-seq bead was annotated with a cell type. The high spatial resolution (10 µm) offered by Slide-seq revealed a detailed view of the cellular architecture that was highly consistent with what one would expect from standard H&E staining. 

In healthy prostate tissues, the spatial data showed well-organized prostate epithelial glandssurrounded by immune and non-immune stromal cells including fibroblasts, pericytes, and endothelial cells. In contrast, the tissue architecture was notably disrupted in tumors.

Adapted from Figure 1 of Hirz et al.: Overall spatial distribution of major cell types revealed from Slide-seq analysis of healthy, adjacent-normal, low-grade tumor (LG) and high-grade tumor (HG) prostate tissues.

Diving deeper into the epithelial cell population,  the data showed a marked decrease in organization and changes in epithelial cell populations as the tumor established and progressed. There was an increase in luminal cells in adjacent and normal tissue but a loss of those luminal cells as the tumor took over. They also observed disruption of the club and hillock cells in tumor tissue.

Adapted from Figure 2 of Hirz et al.: Spatial distribution of epithelial cell types revealed from Slide-seq analysis of healthy, adjacent-normal, low-grade tumor (LG) and high-grade tumor (HG) prostate tissues.

Additionally, spatial mapping of endothelial cell and pericyte populations also showed higher dispersity in tumor tissues as compared with adjacent-normal and healthy tissues. One of the endothelial populations in tumor samples exhibited upregulation of genes in pathways involved in cell migration and proliferation, consistent with their increased dispersity in tumors. The expression pattern of one pericyte population was enriched for pathways in extracellular structure organization and connective tissue development, and the expression pattern of a second pericyte population was enriched for muscle contraction associated with vascular smooth muscle cells. These findings together suggested an increase in endothelial angiogenic activity as the tumor progressed.

Adapted from Figure 3 of Hirz et al.: Spatial distribution of stromal cell types revealed from Slide-seq analysis of healthy, adjacent-normal, low-grade tumor (LG) and high-grade tumor (HG) prostate tissues.

The authors further developed a computational method to infer ligand-receptor interactions between tumor and stromal cells. This analysis was based on observing cooperative upregulation of any ligand-receptor pair in cells that were in close proximity. Given the single-cell scale resolution of Slide-seq, the authors were able to construct a graph of physically adjacent cells and subsequently calculated a ligand-receptor score based on the co-expression level of both ligand and receptor. Among the ~1200 previously curated ligand-receptor pairs, they found 405 significantly statistically significant ligand-receptor interactions.

Adapted from Figure 4 of Hirz et al.: Slide-seq data enabled identification of significant ligand-receptor interactions at tumor-stroma boundary.

Overall, this study demonstrated the utility of Slide-seq in studying the tumor microenvironment and in identifying cell-to-cell communications in the tissue’s native, undissociated state.

Immuno-oncology

Analysis of RCC and melanoma

Mantri et al., PNAS, Dec 2023.

In this Immunity paper, the authors developed Slide-TCR-seq to sequence whole transcriptomes with Slide-seq and TCRs within intact tissues. They were able to show that Slide-TCR-seq maps the characteristic locations of T cells and their receptors in mouse spleen and identified spatially distinct TCR repertoires in human lymphoid germinal centers. Additionally, they also found heterogeneous immune responses in T cells of renal cell carcinoma and melanoma specimens. Here, we summarize a fraction of their findings in mouse lymph nodes and lung metastasis of renal cell carcinoma. 

Figure 1: Comparison of H&E staining or Slide-seq analysis on adjacent sections of mouse spleen. Slide-seq provides structural information concordant to H&E staining as well as whole transcriptome data.

When comparing H&E staining to Slide-seq data, Slide-seq was able to recapitulate the structure of the spleen as well as the H&E staining. In addition to providing the same structural context of H&E staining, Slide-seq also provided the transcriptomes and subsequently the cell location and identity through secondary analysis with RCTD (Fig 1). 

Figure 2: Slide-TCR-seq locates TCR constant and variable region expression in mouse spleen.

When evaluating if Slide-TCR-seq is able to identify T cells, the researchers found that Slide-TCR-seq faithfully recovered both CDR3 sequences, their spatial locations and shows that the constant regions (Fig 2, left side of each pair) and alpha and beta subunits (Fig 2, right side of each pair) of the receptors overlap. 

Once they established that Slide-seq and Slide-TCR-seq were able to identify the correct structures and cell types in normal tissue, the authors looked to apply both techniques to explore the immune environment of tumor tissues. 

Figure 3: Slide-seq and Slide-TCR-seq analysis of a lung metastasis from renal cell carcinoma. A) H&E of adjacent section, B) Slide-Seq analysis defines tumor borders, C) Slide-TCR-seq locates T cell clones with differential gene expression signatures, D) Slide-Seq identifies tumor cell clones, E) Slide-seq identifies the locations of different cell types with RCTD. 

While exploring a lung metastasis of renal cell carcinoma, the authors were able to identify three cell compartments delineating the tumor from lung and intervening boundary (Fig 3B). When looking at the T cell populations within the lung metastasis sample, the authors identified 1,223 unique TCRb T cell clonotypes. Some T cell clonotypes show distinct spatial distributions (Fig 3C). When looking at tumor clonality, the authors were able to identify three different tumor clones with distinct spatial locations (Fig 3D). The authors then applied an annotated single cell reference to the cell populations found in the tumor with RCTD and identified the locations of B cells, dendritic cells, fibroblasts, macrophages, and endothelial cells. This analysis was able to show that immune cells such as B cells, dendritic cells, and macrophages were predominantly located on the tumor edge along the border while the endothelial cells and fibroblasts are more evenly distributed with the tumor (Fig 3E).  

In summary, not only was Slide-seq and Slide-TCR-Seq able to identify known structural and T cell locations, both technologies enabled the authors to identify tumor and T cell clones and differentiate clonal locations within tumor and lymphoid structures.

Cancer development

CANCER DEVELOPMENT

Uncovering spatially restricted gene expression patterns in human melanoma metastasis

Biermann et al., Cell, July 2022.

The dynamic interplay between immune cells and tumor cells strongly influences the development of a tumor and its response to cancer therapy. In this study, Biermann et al. demonstrated the application of Slide-seq to study the interactions of immune and tumor cells in melanoma brain metastasis (MBM) and extracranial melanoma metastasis (ECM).  Importantly, the fine, 10-micron resolution offered by Slide-seq enabled the researchers to resolve down to individual T and NK cells, the smallest immune cells, thereby providing a comprehensive picture of the cellular composition of the tumor microenvironment.

Across multiple patient tumor samples, the authors found multiple small patches of lymphoid aggregates, signified by co-localization of immunoglobulin genes overlapping with co-occurring B and myeloid cells. These lymphoid aggregates were in close proximity to melanoma cells. Such tertiary follicular structures, where immune editing occurs as a result of interactions among tumor cells and immune cells, have previously been found to be strong predictors of response to immunotherapy.

Adapted from Figure 6 of Biermann et al.: Small patches of lymphoid aggregates, signified by immunoglobulin gene expression and co-localization of B and myeloid cells, were found to be scattered across the tumor in multiple melanoma metastasis samples.

Additionally, the authors found sample-specific, spatially restricted gene expression that might have implications for responses to different types of therapy. 

In one tumor sample, the authors found two spatially distinct regions differing in immune response – one that was highly infiltrated with immune cells (top right, signified with MHC-I expression), and the other devoid of immune cells (bottom left, signified by TIMP1 expression). The area that lacked infiltrating immune cells coincided with the expression of TIMP1, an inhibitor of matrix metalloproteinases that digest extracellular matrix and might have contributed to immune evasion. Additionally, regions of MHC-1 expression did not always coincide with chemokine or IFN expression, suggesting that antigen presentation alone was not sufficient to promote anti-tumor immunity.

Adapted from Figure 6 of Biermann et al.: Spatially restricted immune response found in an ECM sample.

In a second patient sample, the authors discovered another spatially restricted expression pattern involving metabolic pathways. There appeared two major subpopulations of tumor cells, one expressing signatures of the oxidative phosphorylation pathway but with low glycolysis signature, and another with the opposite expression pattern in which expression of glycolysis pathway genes was high. These two tumor subpopulations were spatially defined and might be susceptible to different classes of metabolic drugs.

Adapted from Figure 6 of Biermann et al.:  Spatially restricted expression of different metabolic pathways in a melanoma metastasis sample.

In summary, by utilizing high-resolution Slide-seq, the authors were able to identify spatially restricted gene expression patterns that were common across melanoma metastasis samples, as well as spatial patterns that were sample-specific. 


To learn more about this study, listen to this webinar by Dr. Jana Biermann and Dr. Yiping Wang, co-authors of the study.

Immuno-oncology

IMMUNO-oncology

Unveiling spatial heterogeneity of tumor response to anti-PD1 therapy

Wang et al., Nature Genetics, Jan 2023.

In this study, the authors applied Slide-seq to samples taken from patients participating in the first clinical trial using anti-PD-1 antibody MK-3475 (KEYNOTE-001). Specifically, they studied archived frozen samples before treatment and two time points during treatment from a uveal melanoma liver metastasis patient who achieved a partial response.  

The authors identified a clonal signature of tumor cells based on copy number alterations, labeled as ‘Clone 2’, in the sample prior to treatment. This clone was projected onto the spatial maps in the tumor samples in the subsequent two time points and was found to be strongly expanded. In both time points, within the tissue slice analyzed by Slide-seq, there were two contrasting tumor areas identified – one with the immune resistant ‘Clone 2’, with co-expression of immune resistant signature as well as lack of activated immune cells and presence of dysfunctional CD8 T cells, and one with heavy infiltration and immune cell differentiation at the tumor/normal border, suggestive of active immune editing.

Adapted from Extended Data Figure 6 of Wang et al.: Major cell types of tumor samples taken from a patient with uveal melanoma liver metastasis at two time points (+131d, +215d) during anti-PD1 therapy, as revealed by Slide-seq.

Adapted from Extended Data Figure 6 of Wang et al.:  a) Expression of immune resistant signature (ICR) (top row: early and later time point on-treatment) and tumor cell clone 2 (bottom row: early and later time point on-treatment) identified by copy number analysis via single cell and single nuclei RNA-seq experiments. b) Inferred proportion of the two CD8 T cell states (activated and dysfunctional) at the two time points on-treatment.

This study demonstrated there could be spatially restricted anti-tumor response and therapy resistance even within a tumor, furthering our understanding of the heterogeneity in tumor responses to immunotherapies. 

To learn more about this study, view this webinar by Dr. Jana Biermann and Dr. Yiping Wang, co-authors of the study.