Article Figures & Data
Figures
- Figure 1. Unsupervised analysis of single-cell RNA sequencing (scRNA-seq) data on γδ T cells from peripheral blood of healthy adult donors identifies multiple subtypes.
(A) UMAP-based projection of the merged single cell gene expression data of blood derived γδ T cells from three healthy donors. Different clusters are named arbitrarily with c.γδ prefix (to indicate γδ cells from circulation). (B) Overlay of data source on UMAP-based projection of scRNA-seq data. Cells from donor HD4 and HD5 were pooled before performing scRNA-seq and labeled as HD4/5. The number of cells from each dataset is shown above the projection. (C) Labeling of cells positive for TCR delta genes. Overlay of cells that have genes mapping to the TRDC (left), TRDV2 (middle) and TRGV9 (right) gene segments. (D) Quantification of TRDV2 and TRGV9 in each cluster. y-axis shows the per cent of cells within each cluster (x-axis) of the merged data that is positive for TRDV2 (blue) and TRGV9 (red) gene segments. Numbers above the bars show the number of positive cells. (E) UMAP of PBMC cells coloured by the expression of a selected set of markers, LEF1, SOX4, CCR7, IFNG, CCR6, and RORC. Grey indicates zero expression and purple indicates high expression. (F) Scores of curated effector gene sets for IFNγ production, IL17A production, cytotoxicity, adaptive (i.e., antigen presentation on MHC class 1), and innate gene sets in each of the blood γδ T-cell cluster (x-axis). (G) Heat map showing genes (rows) enriched in each of the cluster (columns). Yellow represents enrichment and purple represents depletion. Only the top four genes per cluster are labeled.
- Figure S1. Evaluation of robustness of PBMC data clustering.
(A) Bar plot showing the percentage of cells per post-Canonical Correlation Analysis cluster of the PBMC data in each cell-cycle phase predicted by cyclone1. (B) UMAP of post-Canonical Correlation Analysis clustered PBMC data after removal of cell-cycle and mitochondrial genes. ARI (title) indicates agreement to the original clustering shown in Fig 1. (C) Accuracy of clustering subsamples of the original PBMC data of different proportions of cells (x-axis). [1] Scialdone A, Natarajan KN, Saraiva LR, Proserpio V, Teichmann SA, Stegle O, Marioni JC, Buettner F (2015) Computational assignment of cell-cycle stage from single-cell transcriptome data. Methods. 10.1016/j.ymeth.2015.06.021.
- Figure 2. Validation of the blood γδ T-cell subtypes.
(A) Feature plot showing CD16 and CD28, which are published markers of the δ2 subtype of blood γδ T cells (Ryan et al, 2016). Each dot is a cell. Grey indicates no expression and purple indicates high expression. (B) Scores for published gene signatures of CD16 (white) and CD28 (black) δ2 subtypes in the clusters found in our blood γδ T-cell scRNA-seq data (x-axis). P-values were computed using Wilcoxon signed-rank test. (C) Feature plot showing expression of GPR56 and CXCR6, markers that appear to be mutually exclusive in blood δ2 subtypes. (D) Flow cytometry based validation of novel markers, GPR56 (y-axis) and CXCR6 (x-axis), in peripheral blood γδ T cells TCRδ2 subtypes. Healthy donor identities (which include blood γδ T cells from two new donors, HD9 and HD10) are indicated in the title. Numbers in each quadrant indicate percentage of δ2 cells.
- Figure 3. Characterization of breast tumour-infiltrating γδ T cells uncovers a subtype that is associated with favourable outcome.
(A) UMAP-based projection of the merged single-cell gene expression data of breast tumour-infiltrating immune cell datasets from two patients. Three clusters were double positive for CD3 and TRDC were classified as γδ T cells (CD3E+). Mph, Macrophage (CD14+), T-reg (FOXP3+), regulatory T cells; B, B cells (CD19+); CD8-T, CD8+ αβ T cells; CD4-T, CD4+ αβ T cells. (B) Overlay of donor identity on UMAP-based projection of scRNA-seq data. The number of cells from each donor is shown above the projection. BC1 is a triple-negative subtype and BC2 is a Her2+ subtype of breast cancer (BC). (C) Identification of cells positive for genes encoding TCR δ chain. Overlay of cells that have genes mapping to the TRDC (left) and TRDV2 (right) gene segments. (D) Quantification of enrichment of genes encoding TRDV2 and TRGV9 γ chain in the BC γδ T clusters. y-axis shows the per cent of cells positive for the indicated TRDV2 (blue) and TRGV9 (white) gene within each BC γδ T-cell cluster (x-axis). Numbers above bars show the number of positive cells. (E) UMAP of BC γδ T cells coloured by the expression of a selected set of markers for identification of IFNG and IL17A subtypes. Grey indicates no expression and purple for high expression. (F) Distribution of gene signature scores (y-axis) for IFNγ production, IL17A production, cytotoxicity, adaptive (i.e., antigen presentation on MHC class 1), and innate gene sets in each of the BC γδ T-cell cluster (x-axis). (G) Kaplan–Meir survival curve of the The Cancer Genome Atlas breast cancer data (Ciriello et al, 2015). Patients were partitioned into high and low group based on scores for gene signatures of each of the indicated BC γδ T-cell cluster. y-axis shows overall survival. (H) Heat map showing top differentially expressed genes (row labels) between the three BC γδ T-cell subtypes. Yellow represents high expression and purple represents low expression.
- Figure S2. Specificity of BC extracted γδT-subtype gene signatures.
(A, B, C) Boxplots showing the distribution of scID scores of the BC1 cells within each cluster (x-axis) for the γδT.1 (A), γδT.2 (B), and γδT.3 (C) gene signatures. P-values indicate significance levels of Wilcoxon test between the mean scores of two clusters.
- Figure S3. Predictors associated with better survival are lower in TCGA BRCA patients that score higher for expression of BC cluster 2 gene signature
. (A) Boxplot of expression (y-axis) of CD3, CD4, and CD8 genes in the The Cancer Genome Atlas (TCGA) breast cancer data samples with high (red) and low (blue) expression of the BC γδ-T.2 subtype gene signature (x-axis). (B) Boxplot of mutational load (y-axis) of the TCGA breast cancer data samples with high and low expression of the BC γδ-T.2 subtype gene signature (x-axis). Wilcoxon rank test was used to compute statistical significance of different scores within each cluster. (C) Boxplot of cytolytic score (y-axis) of the TCGA breast cancer data samples with high and low expression of the BC γδ-T.2 subtype gene signature (x-axis). Wilcoxon rank test was used to compute statistical significance of different scores within each cluster.
- Figure 4. Comparison of γδ T cells from peripheral blood and breast tumour.
(A) Comparison of expression of selected genes (rows) involved in activation, cytotoxicity, exhaustion, and naive T-cell state between blood γδ T-cell and breast tumour-infiltrating γδ T-cell subtypes. Grey represents low average expression and red represents high average expression of the genes in each subtype (columns). (B) Assessment of similarity of the γδ T-cell subtypes in blood and breast tumour. Boxplot of scID scores (y-axis) of the BC cluster-specific gene signatures in the blood clusters (x-axis). Scores above the dashed line indicates enrichment of the indicated gene signature. Mann–Whitney U test was used to compute statistical significance of different scores within each cluster. “***” indicates P < 0.001. (C) Feature plots showing expression of suggested cluster defining markers in the γδ T-cell subtypes in blood (top row) and BC (bottom row). Grey indicates low expression and purple indicates high expression. (D) Table summarizing the proposed refinement of subtype classification of γδ T cells supported by the scRNA-seq data from this study.
Supplementary Materials
Supplemental Data 1.
In this Issue
Related Articles
Cited By...
- No citing articles found.