Mapping the immune environment in clear cell renal carcinoma by single-cell genomics

Abstract

Clear cell renal cell carcinoma (ccRCC) is one of the most immunologically distinct tumor types due to high response rate to immunotherapies, despite low tumor mutational burden. To characterize the tumor immune microenvironment of ccRCC, we applied single-cell-RNA sequencing (SCRS) along with T-cell-receptor (TCR) sequencing to map the transcriptomic heterogeneity of 25,688 individual CD45⁺ lymphoid and myeloid cells in matched tumor and blood from three patients with ccRCC. We also included 11,367 immune cells from four other individuals derived from the kidney and peripheral blood to facilitate the identification and assessment of ccRCC-specific differences. There is an overall increase in CD8⁺ T-cell and macrophage populations in tumor-infiltrated immune cells compared to normal renal tissue. We further demonstrate the divergent cell transcriptional states for tumor-infiltrating CD8⁺ T cells and identify a MKI67 + proliferative subpopulation being a potential culprit for the progression of ccRCC. Using the SCRS gene expression, we found preferential prediction of clinical outcomes and pathological diseases by subcluster assignment. With further characterization and functional validation, our findings may reveal certain subpopulations of immune cells amenable to therapeutic intervention.

Fig. 1. Single-cell RNA sequencing results from immune cells in ccRCC.

a UMAP of 37,055 primary immune cells of peripheral blood, normal renal parenchyma and tumor-infiltrating ccRCC patients. b Distribution of cells by tissue type, peripheral blood (blue), tumor (red), and kidney (light blue). Arrows indicated potential enriched or unique immune cells populations for tissue type. c Percent of cells expressing canonical immune cell markers across the UMAP. d Normalized correlation values for predicted immune cell phenotypes based on the SingleR R package for each cluster; dendrogram based on Euclidean distance. e UMAP demonstrating inferred immune cell types in ccRCC, clusters are colored by cell type and proportion of single-cell per sequencing run by tissue type. P values based on one-way ANOVA; lack of labeled p values equates to value >0.05.

Fig. 2. Clonal dynamics vary by T-cell types and patients.

a UMAP of 37,055 primary immune cells overlaid with the frequency of clonotypes assigned by sample identification. b Occupied repertoire space for the indicated clonotype groups for CD8⁺ and CD4⁺ T cells by sample and tissue type in ccRCC patients. c Clonal overlap quantification by overlap coefficient for CD8⁺ and CD4⁺ T cells by sample and tissue type in ccRCC patients. d The top ten clonotypes for each patient as a relative proportion of clonotypes for corresponding peripheral or tumor populations. Each color represents a unique clonotype by patient. e Distribution of clonotypes by tissue, UMAP cluster and ccRCC patient with highlighted (red) the top two clonotypes, comprising tumor-specific clonotypes across all clusters.

Fig. 3. CD8+ T cells in ccRCC tumors exhibit a transcriptional continuum with distinct populations.

a UMAP subclustering of CD8⁺ T cells (original clusters 1, 8, 9, and 17). b UMAP distribution of single cells by tissue type with relative and absolute percent of cells by tissue in each cluster. c Cell-cycle regression assignments for CD8⁺ T cells by cluster assignment. d Percent of cells expressing selected markers for T-cell biology. e CD8⁺ UMAP of clusters (upper panel) and clonotype frequency (lower panel) overlaid with slingshot-based cell trajectory starting at CD8_4 and proceeding into five distinct curves: branch 1 (B₁), B₂, B₃, B₄, and B₅. f Clonotype overlap coefficients between subclusters. g Z-transformed normalized enrichment scores from ssGSEA for selected gene sets by subcluster. h Normalized enrichment scores for therapeutic response or lack of response to anti-PD-1 therapy across the CD8⁺ T cells (upper panel) and by pseudotime of each branch (lower panel).

Fig. 4. Single-cell CD4+ T-cell characterization within ccRCC.

a UMAP subclustering of CD4⁺ T cells (original clusters 4, 6, 10, 13, 15, and 20). b UMAP distribution of single cells by tissue type with relative and absolute percent of cells by tissue in each cluster. c Percent of cells expressing selected markers for T-cell biology. d CD4⁺ UMAP of subclusters (upper panel) and clonotype frequency (lower panel) overlaid with slingshot-based cell trajectory starting from CD4_1 (root 1) and CD4_3 (root 2) with relative pseudotime for all curves calculated using slingshot. e Occupied repertoire space for CD4⁺ subclusters. f Top ten markers for TI-predominant CD4⁺ subclusters. Size of points are relative to percent of cells in the subcluster expressing the indicated mRNA species. g Z-transformed normalized enrichment scores from ssGSEA for selected gene sets by subcluster.

Fig. 5. Single-cell myeloid characterization in ccRCC.

a UMAP subclustering of myeloid cells (original clusters 4, 6, 10, 13, 15, and 20). b UMAP distribution of single cells by tissue type with relative percent of cells by tissue in each cluster. c Percent of cells expressing selected markers for myeloid and macrophage markers. P values derived from one-way ANOVA testing. d Proportion of assigned cell types compared to total antigen presenting cells by tissue type. e Macrophage subclusters: tumor-associated macrophage 1 (TAM_1) (n = 1262), TAM_2 (n = 840), TAM_3 (n = 594), peripheral macrophage (pM) (n = 275), and resident macrophage (rM) (n = 194) with relative and absolute percent of cells by tissue in each cluster. f Top differential expression markers for macrophage subclusters. g Macrophage UMAP overlaid with slingshot-based cell trajectories starting at rM and TAM_2 and proceeding into pM. Smaller UMAP shows pseudotime created by the cell trajectories. h Z-transformed normalized enrichment scores from ssGSEA for selected gene sets by subcluster.

Fig. 6. Prognostic values of gene signatures derived from CD8+ T cell and TAM subclusters.

a Schematic diagram of the machine-learning approach for signature development, selection and testing based on the k-nearest neighbors algorithm using the TCGA renal clear cell carcinoma data set. b Kaplan–Meier curves for overall survival in testing subset for the CD8_6 subcluster signature with corresponding distribution of histological grades by model assignment. c Kaplan–Meier curves for overall survival in testing subset for the TAM_3 subcluster signature with corresponding distribution of histological grades by model assignment with logrank p values. d Density plots for 937,713 T cells isolated from four healthy samples, 68 primary ccRCC, four metastasis, and quantified using mass cytometry. e Markers with increased median expression in CD45⁺ CD3⁺ CD8⁺ PD-1⁺ Ki-67⁺ cells compared to other CD8⁺ cells. Adjusted p values < 1e−12 for all indicated markers using Welch’s T test. f ccRCC primary tumor and healthy samples subdivided into tertiles by the proportion of CD45⁺ CD3⁺ CD8⁺ PD-1⁺ Ki^-67⁺ relative to the entire CD45⁺ CD3⁺ CD8⁺ pool by histological grade.