Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 May 19;31(7):107628.
doi: 10.1016/j.celrep.2020.107628.

Single-Cell RNA Sequencing Reveals a Dynamic Stromal Niche That Supports Tumor Growth

Affiliations
Free PMC article

Single-Cell RNA Sequencing Reveals a Dynamic Stromal Niche That Supports Tumor Growth

Sarah Davidson et al. Cell Rep. .
Free PMC article

Abstract

Here, using single-cell RNA sequencing, we examine the stromal compartment in murine melanoma and draining lymph nodes (LNs) at points across tumor development, providing data at http://www.teichlab.org/data/. Naive lymphocytes from LNs undergo activation and clonal expansion within the tumor, before PD1 and Lag3 expression, while tumor-associated myeloid cells promote the formation of a suppressive niche. We identify three temporally distinct stromal populations displaying unique functional signatures, conserved across mouse and human tumors. Whereas "immune" stromal cells are observed in early tumors, "contractile" cells become more prevalent at later time points. Complement component C3 is specifically expressed in the immune population. Its cleavage product C3a supports the recruitment of C3aR+ macrophages, and perturbation of C3a and C3aR disrupts immune infiltration, slowing tumor growth. Our results highlight the power of scRNA-seq to identify complex interplays and increase stromal diversity as a tumor develops, revealing that stromal cells acquire the capacity to modulate immune landscapes from early disease.

Keywords: cancer-associated fibroblast; cell-cell communication; immune; melanoma; single-cell sequencing; stroma; tumour microenvironment.

Conflict of interest statement

Declaration of Interests The authors declare no competing interests.

Figures

None
Figure 1
Figure 1
Distinction of Melanoma Stromal Populations with Single-Cell RNA-Seq (A) Overview of experimental and sequencing workflow. (B) t-Distributed Stochastic Neighbor Embedding (tSNE) visualization of all cells sequenced with each cell color coded for cell type (left), site of origin (center), and time (right). (C) Expression of marker genes for each cell type. n = 32 mice. cDC1/2, conventional dendritic cell; DC LN, lymph node dendritic cell; Endo lymph, lymphatic endothelial cell; endo LN, lymph node endothelium; Endo tumor, tumor endothelial cells; fibroblast LN, lymph node fibroblast; MAIT, mucosal-associated invariant T cell; migDC, migratory DC; NK, natural killer; pDC, plasmacytoid DC.
Figure 2
Figure 2
Myeloid Cell Clusters in the Tumor Exhibit Suppressive Characteristics (A) tSNE plot of individual myeloid cells colored by site (tumor, dark gray; lymph node, light gray) and clusters marked by colored lines. (B) tSNE plots showing the expression of selected marker genes for macrophages and inflammatory and resident monocytes. (C) Violin plots showing the expression of selected surface marker genes within each cell cluster displayed as log (TPM+1). TPM, transcript count per million. (D) Heatmap showing mean expression (log(TPM+1)) of co-stimulatory and suppressive genes for the identified cell clusters. (E) Heatmap showing the relative expression (Z score) of co-stimulatory and suppressive genes in all innate immune cells over time. (F) Flow cytometric analysis of tumor infiltrating CD11b+ cells for the expression of suppressive markers PDL1 and Arg 1 at days 6 and 11. Data presented as means ± SEMs; day 6 n = 12 independent mice and day 11 n = 11 independent mice. ∗∗∗∗p < 0.0001 (t test). (G) Schematic diagram of the co-stimulatory and inhibitory receptors-ligands expressed on distinct myeloid subpopulations. For (A)–(E) and (G), n = 17 mice. cDC1/2, conventional dendritic cell; pDC, DC LN, lymph node dendritic cell; migDC, migratory DC; MP, mononuclear phagocyte; plasmacytoid DC.
Figure 3
Figure 3
T Cells Recruited from Lymph Nodes Are Activated In Situ (A) tSNE plot of individual T cells colored by site (tumor, dark gray; lymph node, light gray) and annotated subpopulations marked by colored lines. (B) Heatmap showing relative expression (Z score) of functional gene groups for cell clusters. (C) Pseudotime analysis of CD8+ T cell gene trajectories colored by site (left), clonal expansion (center), and tumor stage (days, right); arrow indicates time direction. (D) Expression of activation-associated genes along the inferred pseudotime colored by site; lymph node (green), tumor (blue). (E) Flow cytometric analysis of T cells isolated from skin and day 5 and 11 tumors, as well as their draining lymph nodes. The number of CD8+ cells was quantified, as was proliferation (Ki67) and PD1 expression. Data presented as means ± SEMs, n = 4 independent mice for each condition. p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 (two-way ANOVA with a Sidak post hoc test). For (A)–(D), n = 10 mice.
Figure 4
Figure 4
Distinct Fibroblast Clusters Identified in Melanoma Tumors (A) tSNE plot of sequenced CD31 stromal cells from tumors colored by their associated cluster. (B) Heatmap showing average expression (log(TPM+1)) of typical mesenchymal markers. (C) Heatmap of Gene Ontology (GO) pathways for differentially expressed genes in each cluster, including cytokine-chemokine receptor interactions, complement cascade, extracellular matrix interactions, and actin cytoskeleton. Columns represent individual cells and rows display Z scores. (D) Sequencing data represented as a bar plot, depicting the ratio of stromal populations at each time point examined. The size of each colored bar is proportional to the percentage of total stromal cells each population represents. Data presented as means ± SEMs, n = 7 mice. p < 0.05 (two-way ANOVA with Tukey post hoc test). (E) tSNE plot of sequenced fibroblasts from tumors by tumor time point (right). (F) tSNE visualization of the proliferation marker Mki67 in the CAFs. (G) Heatmap depicting logistic regression analysis of normal mouse skin, indicating to which of the 3 stromal clusters these cells are most similar. scRNA-seq of melanoma samples, n = 7 mice. scRNA-seq of healthy murine skin samples, n = 2 mice. scRNA-seq of healthy human skin, n = 1 sample.
Figure 5
Figure 5
Conservation of Fibroblast Subpopulations between Murine Tumor Types (A) Representative confocal images of PDPN, PDGFRα, and αSMA in combination, or CD34 in combination with either PDPN, PDGFRα, or αSMA (right panel) in day 5 and day 11 tumors. Dashed line indicates the tumor border. Scale bars, 100 μm; images represent at least n = 3 independent mice. (B) Flow cytometry quantification of the proportion of each stromal population at day 5 and day 11 tumors, displayed as a percentage of the total stromal population. Skin, n = 8 mice; day 5, n = 25 mice; day 11, n = 30 mice. (C) Flow cytometric quantification of intracellular CXCL12 and C3 expression in each population presented as fold change in mean fluorescence at day 11, normalized to the CD34high αSMAlowpopulation. CXCL12, n = 42 tumors; C3, n = 12 tumors. (D) Representative confocal images of CSF1 expression in CD34+ stromal populations in day 5 and day 11 tumors. Scale bars, 50 μm; images represent at least n = 2 independent mice. (E) Schematic diagram of the 3 stromal subpopulations. (F) Representative confocal images of stromal population markers in orthotopic E0771 breast tumors. Representative confocal images of PDPN, PDGFRα, and αSMA in combination (top panel) or CD34 in combination with either PDPN, PDGFRα, or αSMA (bottom panel). Dashed line indicates the tumor border. The asterisk indicates colocalization between CD34 and PDPN or PDGFRα; arrowhead indicates CD34 expression that is distinct from PDPN or PDGFRα. Scale bars, 100 μm; images represent at least n = 3 independent mice. (G) Flow cytometric quantification of intracellular C3 expression in each E0771 breast stromal population presented as fold change in mean fluorescence at day 16 normalized to the CD34high αSMAlow. n = 8 independent mice. Each point represents a tumor. Data presented as means ± SEMs. p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; one-way ANOVA with a Tukey post hoc test.
Figure 6
Figure 6
Stromal-Immune Crosstalk Supports the Development of an Immunosuppressive Niche (A) Overview of selected statistically significant interactions between stromal subsets and other cell types using a cell-cell communication pipeline based on CellPhoneDB. Size indicates p values (permutation test, see STAR Methods), and color indicates the means of the receptor-ligand pairs between 2 clusters. (B) Violin plots displaying expression log(TPM+1) of ligands Cxcl12, Csf1, and C3 and cognate receptors Cxcr4, Csf1r, and C3ar1 on respective stromal populations. n = 26 mice. (C) Confocal images of representative tumor-tissue borders. CXCR4, CSFR1, or C3aR expressing macrophages located proximally to CD34+ CAFs (green, F4/80; red, CXCR4, CSF1R, or C3aR; white, PDPN; blue, CD34). Scale bars, 50 μm. (D) Flow cytometric quantification of CXCL12 and C3 expression across compartments of the tumor microenvironment. Each point represents a tumor. CXCL12 n = 42 tumors, C3 n = 12 tumors. One-way ANOVA with Tukey post hoc test. (E) In vivo blockade of C3a in established tumors. Top left: experimental design and treatment regimen; top right: tumor volume (in cubic millimeters) of mice treated with IgG control (blue) or anti-C3a (red); bottom left: myeloid infiltration in day 6 tumors, after 24 h of treatment with IgG or anti-C3a. The number of F4/80 and Ly6C+ Ly6G cells are shown as a percentage of Cd11b and CD45 cells, respectively; bottom right: the number of tumor-infiltrating CD8+ T cells at day 11, displayed as raw counts normalized to tumor volume (in cubic millimeters). Data presented as means ± SEMs. n = minimum 13 mice. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; t test. (F) Schematic diagram of the dynamic crosstalk identified within the tumor microenvironment.
Figure 7
Figure 7
Similar Stromal Populations and C3-C3aR Interactions Are Conserved in Human Melanoma and Head and Neck Cancer Publically available single-cell sequencing data from human melanoma and head and neck cancer were downloaded and analyzed. (A) tSNE plots of sequenced populations for melanoma. (B) Heatmap depicting stromal subsets 1–3 defined by similar markers and functional features to murine melanoma dataset. Heatmap displays the expression (Z scores, blue to red) of key markers and cytokines across stromal clusters identified in human melanoma. (C) Overview of statistically significant interactions between stromal subsets and other cell types using the CellPhoneDB pipeline. Size indicates p values and color indicates the means of the receptor-ligand pairs between 2 clusters. (D) Violin plots displaying conserved expression log(TPM+1) of C3 and cognate receptor C3ar1 on respective stromal populations in human melanoma. n = 19 patient samples. (E) tSNE plots of sequenced populations for human head and neck cancer. (F) Heatmap depicting stromal subsets 1–3 defined by similar markers and functional features to murine melanoma dataset. Heatmap displays expression (Z scores, blue to red) of key markers and cytokines across stromal clusters identified in human head and neck cancer. (G) Statistically significant interactions between stromal subsets and other cell types using the CellPhoneDB pipeline. Size indicates p values and color indicates the means of the receptor-ligand pairs between 2 clusters. (H) Violin plots displaying conserved expression log(TPM+1) of C3 and cognate receptor C3ar1 on stromal populations in human head and neck cancer. n = 18 patient samples.

Similar articles

See all similar articles

References

    1. Ajona D., Ortiz-Espinosa S., Moreno H., Lozano T., Pajares M.J., Agorreta J., Bértolo C., Lasarte J.J., Vicent S., Hoehlig K. A Combined PD-1/C5a Blockade Synergistically Protects against Lung Cancer Growth and Metastasis. Cancer Discov. 2017;7:694–703. - PubMed
    1. Angell H.K., Bruni D., Barrett J.C., Herbst R., Galon J. The Immunoscore: Colon Cancer and Beyond. Clin. Cancer Res. 2020;26:332–339. - PubMed
    1. Arora P.D., Narani N., McCulloch C.A. The compliance of collagen gels regulates transforming growth factor-beta induction of alpha-smooth muscle actin in fibroblasts. Am. J. Pathol. 1999;154:871–882. - PMC - PubMed
    1. Augsten M. Cancer-associated fibroblasts as another polarized cell type of the tumor microenvironment. Front. Oncol. 2014;4:62. - PMC - PubMed
    1. Bartoschek M., Oskolkov N., Bocci M., Lövrot J., Larsson C., Sommarin M., Madsen C.D., Lindgren D., Pekar G., Karlsson G. Spatially and functionally distinct subclasses of breast cancer-associated fibroblasts revealed by single cell RNA sequencing. Nat. Commun. 2018;9:5150. - PMC - PubMed

LinkOut - more resources

Feedback