Single-cell sequencing links multiregional immune landscapes and tissue-resident T cells in ccRCC to tumor topology and therapy efficacy

Abstract

Clear cell renal cell carcinomas (ccRCCs) are highly immune infiltrated, but the effect of immune heterogeneity on clinical outcome in ccRCC has not been fully characterized. Here we perform paired single-cell RNA (scRNA) and T cell receptor (TCR) sequencing of 167,283 cells from multiple tumor regions, lymph node, normal kidney, and peripheral blood of two immune checkpoint blockade (ICB)-naïve and four ICB-treated patients to map the ccRCC immune landscape. We detect extensive heterogeneity within and between patients, with enrichment of CD8A⁺ tissue-resident T cells in a patient responsive to ICB and tumor-associated macrophages (TAMs) in a resistant patient. A TCR trajectory framework suggests distinct T cell differentiation pathways between patients responding and resistant to ICB. Finally, scRNA-derived signatures of tissue-resident T cells and TAMs are associated with response to ICB and targeted therapies across multiple independent cohorts. Our study establishes a multimodal interrogation of the cellular programs underlying therapeutic efficacy in ccRCC.

Keywords: TCR sequencing; immunotherapy; pathology; renal cell carcinoma; single-cell RNA-sequencing; tissue-resident; tumor-associated macrophages; tyrosine kinase inhibitors.

Figure 1.. Patient characteristics and experimental design

(A) Patients in the study. Numbers on each patient represent age. Baseline demographic characteristics as well as management course are displayed. (B) Representative computed-tomography (CT) images obtained during patient management. For individuals who were treated with immune checkpoint blockade (ICB), disease sites are shown before and after treatment administration. (C) Schematic of the experimental design used in this study. Multiple regions from renal cell carcinoma (RCC) tumors, normal kidney, and peripheral blood mononuclear cells (PBMCs) were profiled. (D–F) Representative hematoxylin/eosin (H&E) images. (D) H&E image (magnification 40x, scale bar represents 250 μm) of a tumor region showing complete tumor regression with extensive hyalinization and fibrosis. (E) Representative tumor region showing a peritumoral lymphocytic infiltrate (magnification 100x, scale bar represents 250 μm). In this particular case, the sample showed extensive tumor regression (fibrosis/hyalinization) and a moderate perivascular/peritumoral infiltrate in the viable areas. (F) Representative H&E image (magnification 40x, scale bar represents 20 μm) of a tumor sample showing minimal immune infiltration. See also Figure S1.

Figure 2.. Immune landscape of patients with ccRCC at single cell resolution

(A) Uniform manifold approximation and projection (UMAP) embedding of transcriptional profiles from all patients and samples (n = 167283). Each dot represents a single cell, and colors represent clusters denoted by inferred cell type. (B) Normalized expression of markers for immune cells (CD45), T cells (CD3D), myeloid cells (CD14), and ccRCC cells (CA9). For all UMAP plots, yellow indicates high log-normalized expression; blue indicates low. (C) UMAP embedding of single cell transcriptional profiles from each tissue sampled in the study. Each plot includes cells from all patients for whom the tissue was sampled. (D) Normalized expression of common markers for lymphoid (T, NK, and B) populations. (E) Normalized expression of differentially expressed genes between all CD8⁺ exhausted T cell clusters (top row) and CD4⁺ T cell clusters (bottom row). Violin plot top and bottom lines indicate range of normalized expression; width indicates number of cells at the indicated expression level. (F) Normalized expression of common markers for myeloid (macrophage, monocyte, dendritic cell, and mast) populations. (G) Normalized expression of differentially expressed genes between all tumor-associated macrophage (TAM) clusters. Violin plot top and bottom lines indicate range of normalized expression; width indicates number of cells at the indicated expression level. (H) Correlations between CLR-transformed prevalences of representative clusters across all tumor regions (n = 18) from all patients. P-values calculated using Spearman correlation. See also Figures S2, S3, S4, and Table S1.

Figure 3.. Multiregional immune heterogeneity across ccRCC patients

(A) Prevalence of each single cell cluster (columns) across all samples (n = 29) from all patients studied (rows). Purple dots indicate tumor samples, brown dots indicate normal kidney samples, red indicate PBMC samples, and gray dots indicate lymph node samples. For clarity, cluster prevalences < 1% are left blank. (B) Representative H&E images for tumor regions show in (A). 1, magnification 40x, scale bar represents 8 μm; 2, magnification 40x, scale bar represents 20 μm; 3, magnification 40x, scale bar represents 250 μm ; 4, magnification 100x; scale bar represents 20 μm; 5, magnification 100x; scale bar represents 200 μm; 6, magnification 40x; scale bar represents 250 μm. (C) Expression of the Javelin signature across all samples (n = 29) in (A). Boxplots show first and third quartiles and median; bottom whiskers represent range from minimum to first quartile; top whiskers indicate range from third quartile to maximum. (D) Z-scored ssGSEA scores for signatures of scRNA clusters, applied to tumor samples assesed via bulk RNA-sequencing. Each column of the heatmap represents a tumor region from a single patient (n = 29 patients total; 131 samples total). Column annotations show tumor regions from 9 representative patients (3 untreated, 3 treated with ICB, and 3 treated with combination TKI + ICB). (E) Z-scored ssGSEA scores for signatures of scRNA clusters, applied to normal kidney samples assesed via bulk RNA-sequencing. Each column of the heatmap represents a normal kidney region from a single patient (n = 29 patients; 69 samples total). Column annotations show normal kidney regions from 9 representative patients (3 untreated, 3 treated with ICB, and 3 treated with combination TKI + ICB). See also Figures S5 and S6.

Figure 4.. Phenotype of expanded intratumoral TCR clonotypes, TCR trajectory analysis, and TCR clonotype sequence similarity

(A) Phenotypes of top 50 most frequent intratumoral TCR clonotypes (Vα, Vβ , Jα, Jβ , CDR3α, and CDR3β) in each patient. Rows of each heatmap are unique clonotypes. Heatmaps show log2 count of each clonotype, where the count represents the number of times the clonotype is observed in the tumor. Columns (color bars) indicate scRNA clusters from which each clonotype originates. (B) Force-directed layout incorporating all T cells sharing TCR clonotypes with the tissue-resident cluster. All 6 patients are included in the analysis and embedded in the same space. (C) Embedding from (B) colored by patient. Patients with most number of cells are labeled in black. (D) Pseudotime ordering of cells starting in the tissue-resident cluster. (E) Zoomed view of (B) showing cells from the complete responder. Data show a trajectory between tissue-resident T cells and NK-like cells. (F) Zoomed view of (B) showing cells from the resistant patient. Data show no trajectory between tissue-resident T cells and NK-like cells. (G) Force-directed layout incorporating all T cells sharing TCR clonotypes with the CD8⁺ NK-like cluster. All 6 patients are included in the analysis and embedded in the same space. (H) Embedding from (G) colored by patient. Patients with most number of cells are labeled in black. (I) Pseudotime ordering of cells starting in the NK-like cluster. (J) Zoomed view of (H) showing cells from the complete responder. Data show a trajectory between NK-like cells and tissue-resident T cells. (K) Zoomed view of (H) showing cells from the resistant patient. Data show no trajectory between NK-like cells and tissue-resident T cells. (L–N) Pairwise similarity (Grantham distance) between the top 50 most frequent intratumoral TCR CDR3β clones. Similarity was assessed between CDR3β amino acid sequences of the most frequent length in each patient. For (L–M), sequence logos show consensus sequence for heatmap regions of high similarity, surrounded by green boxes. Sequences to right of each logo are the CDR3β sequences in each green box, and red indicates the most expanded clone. For (N), the sequence of the most expanded clone is shown. See also Figure S7.

Figure 5.. Immune populations underlying clinical gene signature performance

(A) Heatmap of mean log-normalized expression of clinical signature genes across all immune and non-immune scRNA clusters, standardized by z-score. (B–E) Association of clinical signatures with survival across the TCGA ccRCC cohort and the COMPARZ cohort. Squares and lines indicate hazard ratios and 95% confidence intervals, respectively. Hazard ratios calculated using univariable Cox regression; P-values calculated using log-rank test. (F) Association of clinical signatures with response and resistance to ICB in the IMmotion 150 cohort. (G) Association of clinical signatures with response and resistance to ICB+TKI in the IMmotion 150 cohort. P-values calculated using two-sided Wilcoxon test. Blue stars indicate significance at FDR P < 0.05. See also Figure S8.

Figure 6.. Single cell-derived signatures of CD8A⁺ tissue-resident T cells and TAM ISG^hi cells predict improved PFS after ICB-based combination regimens and anti-VEGF therapy

(A) Application of the CD8A⁺ tissue-resident signature to pre-therapy bulk RNA-seq data from the Atezo/Bev arm of the Immotion 151 cohort. Each p-value is from a multivariable Cox regression analysis testing the given signature and controlling for MSK risk score. (B) Application of the TAM ISG^hi signature to pre-therapy bulk RNA-seq data from the Sunitinib arm of the IMmotion 151 cohort. Each p-value is from a multivariable Cox regression analysis testing the given signature controlling for MSK risk score and sarcomatoid status (yes/no). (C) Application of the CD8A⁺ tissue-resident signature to pre-therapy bulk RNA-seq data from the Avelumab/Axitinib arm of the Javelin 101 cohort. (D) Application of the TAM ISG^hi signature to pre-therapy bulk RNA-seq data from the Sunitinib arm of the Javelin 101 cohort. See also Figure S9 and tables S2 and S3.