Single-cell RNA sequencing reveals a dynamic stromal niche within the evolving tumour microenvironment

Abstract

Non-cancerous stromal cells represent a highly diverse compartment of the tumour, yet their role across tumour evolution remains unclear. We employed single-cell RNA sequencing to determine stromal adaptations in murine melanoma at different points of tumour development. Naive lymphocytes recruited from lymph nodes underwent activation and clonal expansion within the tumour, prior to PD1 and Lag3 expression, while tumourassociated myeloid cells promoted the formation of a suppressive niche through cytokine secretion and inhibitory T cell interactions. We identified three temporally distinct cancerassociated fibroblast (CAF) populations displaying unique signatures, and verified these in human datasets. In early tumours, immune CXCL12/CSF1 and complement -expressing CAFs supported recruitment of macrophages, whereas contractile CAFs became more prevalent in later tumours. This study highlights the complex interplay and increasing diversity among cells that co-evolve with the tumour, indicating that from early stages of development, stromal cells acquire the capacity to modulate the immune landscape towards suppression.In particular, extensive heterogeneity has been reported within tumour fibroblast populations.Cancer associated fibroblasts (CAFs) are the most abundant non-immune stromal component, secreting growth factors, promoting angiogenesis, facilitating metastasis and regulating immune infiltrates 9-15 . Although they express typical fibroblast markers, such as fibroblast activation protein (FAP), platelet derived growth factor receptors α (PDGFRα) and β (PDGFRβ), podoplanin (PDPN), Thy-1 and α-smooth muscle actin (αSMA), no single marker universally identifies all CAFs within the tumour stroma 16-18 . To date, many studies rely on positive selection approaches, in which one or two markers are used to isolate CAFs for functional characterisation. Consequently, these findings likely reflect a sub-population of cells and may bias our perceptions of CAF function. It remains unclear whether fibroblast subpopulations with distinct roles are present in the tumour microenvironment.Current approaches lack the resolution to visualise the true extent of stromal heterogeneity and may mask rare populations, or cellular phenotypes, that could be critical for tumour survival. Therefore, we have employed single-cell RNA sequencing (scRNAseq) to profile both immune and non-immune stromal populations from the B16-F10 murine model of melanoma. Furthermore, cells were isolated from both primary tumours and draining lymph nodes, at different stages of tumour development, enabling a systems level interrogation of the melanoma microenvironment in real-time. Here, we identified the presence of a diverse immune landscape, in which the composition and phenotype of leukocytes change as the tumour evolves. In particular, effector T cells displaying signs of dysfunction, were detected predominantly in late stage tumours. This work also highlighted significant heterogeneity within the CAF compartment of the primary tumour. Three distinc...

Single cell transcriptomics reveals involution mimicry during the specification of the basal breast cancer subtype

“…Functional annotation using GSVA analysis (SuppFig. 7A) revealed that secretory FC0 were governed by hallmarks of EMT and were enriched for a desmoplastic signature 37 , characterised by genes involved in extra cellular matrix (ECM)…”

“…5C and D). Secretory FC1 clearly showed immune-related functions (top enriched hallmarks "Interferon alpha response", "IL6/JAK/STAT3 signalling" and "Interferon gamma response"), with genes involved in cytokine-cytokine receptor interactions, monocyte recruitment and the complement cascade 37 . FC2 were enriched for a contractile signature, characterised by the expression of genes involved in actin cytoskeleton organisation.…”

CellPhoneDB v2.0: Inferring cell-cell communication from combined expression of multi-subunit receptor-ligand complexes

“…For example, CellPhoneDB helped us identify a shift in the cellular communication from a network that was dominated by mesenchymal-epithelial interactions in healthy airways, to a Th2 cell--dominated interactome in asthmatic airways 2 . In another recent 40 example, our cell-cell communication framework revealed the complex interplay among diverse cells in the evolving tumor microenvironment of a murine melanoma model where multiple immunosuppressive mechanisms coexist within a heterogeneous stromal compartment 3 . We anticipate that the information stored in CellPhoneDB 2.0 will have the potential to resolve cellular communication when combined with the spatial location of the cells that can be quantified using highly multiplexed spatial methods [4][5][6][7] .…”

Northstar enables automatic classification of known and novel cell types from tumor samples

“…To test northstar's performance on a larger cancer dataset, we analyzed 4,626 cells from a melanoma murine model focused on infiltrating immune cells 21 . As with the GBM data, we found high accuracy with the default parameters (~ 90% correctly assigned immune cells using Tabula Muris marrow annotations only).…”