Single-cell technologies, such as flow cytometry and single-cell RNA-sequencing, are uniquely suited to resolve the cellular diversity in complex tissues such as sarcoma ecosystems. At the same time, they require a single-cell suspension as starting material, which has to be collected from tissue directly after surgical tumor resection. Applying these technologies to study sarcoma ecosystems necessitates the design of sarcoma-tailored experimental protocols. These protocols have to consider the extensive heterogeneity of sarcoma subtypes as well as their origin from diverse soft tissues and bones in the human body. To enable a comprehensive characterization of the cellular diversity in sarcoma tumor ecosystems at the single-cell level in Objective 1, we thus designed a pan-sarcoma tissue dissociation protocol and collected samples from over 200 patients spanning 39 subtypes under written informed consent and ethics approval. To make the most out of this unique single-cell suspension cohort, we designed a high-throughput spectral flow cytometry approach in collaboration with researchers at the Berlin Institute of Health. For this, we designed a cocktail of 36 different antibodies that stain individual proteins at the cell surface, which tell us about each cell’s identity in the sarcoma ecosystem. We focused particularly on the tumor-associated immune cells. With this approach, we were able to detect the abundance levels of 24 distinct immune cell populations among 19 million immune cells from 118 tumors, 50 non-cancerous tissues, and 20 blood samples. That is a data set of unprecedented size to characterize sarcoma ecosystems at the single-cell level. Among the sarcoma-associated immune cells were mainly T cells and macrophages. Less abundant were B cells, natural killer cells, dendritic cells, granulocytes, and monocytes. Each of these immune cell populations may exert different functions in the sarcoma ecosystem, which we want to explore further. We also discovered that the tumor immune composition stratified our patients into six groups, all heterogeneous for sarcoma subtype and tumor grade, which is a marker of tumor aggressiveness. It is a remarkable finding that sarcomas of very different subtypes can exhibit similar immune environments, as this indicates conserved patterns of immune organization across patients and subtypes. These patterns could represent cellular dependencies or vulnerabilities, which demands further investigation. To better understand the consequences of these similarities for sarcoma ecosystem biology, we selected samples from 70 patients guided by our single-cell immune composition map for cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq). This is a combined single-cell RNA sequencing approach that allows i) to assess each cell’s transcriptome as a readout for what the cell is currently doing, ii) to characterize the presence and abundance of 130 surface proteins as a readout for who in the ecosystem the cell might interact with, and iii) to determine which and how many T cells in the ecosystem have recognized an antigen, which is an indicator of T cell-mediated cancer cell killing. We chose 13 different sarcoma subtypes for this experiment. We detected a high abundance of activated cytotoxic CD8 T cells in selected sarcomas of different subtypes. T cell receptor sequencing revealed clonal T cell expansion, which indicates an ongoing anti-tumor immune response in these samples and makes them potential candidates for immune checkpoint inhibition therapy. The novelty of our finding is the detection of activated cytotoxic T cells in less expected sarcoma subtypes, which demands further investigation. Analyses are ongoing to describe these and additional tumor-immune interactions in more detail. In conclusion, while some goals for Objective 1 have only partly been achieved, we by far exceeded our expectations regarding the number of samples and mesenchymal tumor entities for which we were able to create a detailed immune composition map at the single-cell level.
To characterize the spatial organization of immune cell populations within sarcoma ecosystems in Objective 2, we first designed a manual sequential immunofluorescence approach using off-the-shelf antibodies and a conventional 5-color-slide scanner. We successfully validated and titrated 30 different antibodies to detect and characterize immune cell phenotypic diversity in situ. We also expanded our antibody panels to include 9 sarcoma entity-specific markers. Pilot experiments to test the workflow capacity for analyzing our target number of 20 formalin-fixed and paraffin-embedded (FFPE) sections for each of 7 sarcoma entities showed poor feasibility with regard to handling time and ultimate data quality. The introduction of a new multiplexed imaging platform enabled us to translate our antibody panel to this more automated workflow. We were able to acquire 36-plex images in overnight runs with automated image registration and background subtraction, which enabled faster, more efficient, and better image quality results with the downside of higher costs per sample. We, therefore, decided to focus on specific sarcoma samples and successfully selected multiple tissue sections to confirm the findings of Objective 1. As a next step, our workflow will be expanded to tissue microarrays to confirm our findings on larger and more balanced cohorts with regard to entity, outcome, and additional disease parameters. In conclusion, we have reached the most important goals of Objective 2 to set up an imaging workflow and study the spatial organization of the immune landscape in sarcoma ecosystems and have exceeded our expectations with regard to image multiplexity and quality.
A functional assessment of cellular interactions within sarcoma ecosystems in vitro, as aimed for in Objective 3, is outstanding. The main reason is the ongoing extensive analysis of data produced as part of Objectives 1 and 2 to extract cellular relationships and potential vulnerabilities to follow up on in vitro. At the same time, we aimed at reaching aspects of the goal in Objective 3, “to provide insights into how the sarcoma-associated immune system shapes sarcoma tumor cell functional phenotypes”, by including two disease time points as well as samples obtained before and after radiotherapy in our single-cell profiling cohorts. These sample pairs allowed us to better understand differences in tumor cell phenotypes of the same lesion within the context of changes in the immune microenvironment.
