Cancer cells are embedded in the tumor microenvironment (TME), a complex ecosystem of stromal cells. Here, we present a 52,698-cell catalog of the TME transcriptome in human lung tumors at single-cell resolution, validated in independent samples where 40,250 additional cells were sequenced. By comparing with matching non-malignant lung samples, we reveal a highly complex TME that profoundly molds stromal cells. We identify 52 stromal cell subtypes, including novel subpopulations in cell types hitherto considered to be homogeneous, as well as transcription factors underlying their heterogeneity. For instance, we discover fibroblasts expressing different collagen sets, endothelial cells downregulating immune cell homing and genes coregulated with established immune checkpoint transcripts and correlating with T-cell activity. By assessing marker genes for these cell subtypes in bulk RNA-sequencing data from 1,572 patients, we illustrate how these correlate with survival, while immunohistochemistry for selected markers validates them as separate cellular entities in an independent series of lung tumors. Hence, in providing a comprehensive catalog of stromal cells types and by characterizing their phenotype and co-optive behavior, this resource provides deeper insights into lung cancer biology that will be helpful in advancing lung cancer diagnosis and therapy.

The stromal compartment of the tumor microenvironment consists of a heterogeneous set of tissue-resident and tumor-infiltrating cells, which are profoundly moulded by cancer cells. An outstanding question is to what extent this heterogeneity is similar between cancers affecting different organs. Here, we profile 233,591 single cells from patients with lung, colorectal, ovary and breast cancer ( n = 36) and construct a pan-cancer blueprint of stromal cell heterogeneity using different single-cell RNA and protein-based technologies. We identify 68 stromal cell populations, of which 46 are shared between cancer types and 22 are unique. We also characterise each population phenotypically by highlighting its marker genes, transcription factors, metabolic activities and tissue-specific expression differences. Resident cell types are characterised by substantial tissue specificity, while tumor-infiltrating cell types are largely shared across cancer types. Finally, by applying the blueprint to melanoma tumors treated with checkpoint immunotherapy and identifying a naïve CD4 + T-cell phenotype predictive of response to checkpoint immunotherapy, we illustrate how it can serve as a guide to interpret scRNA-seq data. In conclusion, by providing a comprehensive blueprint through an interactive web server, we generate the first panoramic view on the shared complexity of stromal cells in different cancers.

Immune-checkpoint blockade (ICB) combined with neoadjuvant chemotherapy improves pathological complete response in breast cancer (BC). To understand why only a subset of tumors respond to ICB, patients with hormone receptor-positive or triple-negative BC were treated with anti-PD1 prior to surgery. Paired pre-versus on-treatment biopsies from treatment-naïve patients receiving anti-PD-1 (n=29) or patients receiving neoadjuvant chemotherapy prior to anti-PD1 (n=11) were subjected to single-cell transcriptome, T-cell receptor and proteome profiling. One-third of tumors contained PD1-expressing T-cells, which clonally expanded upon anti-PD1 treatment irrespective of tumor subtype. Expansion mainly involved CD8 + T-cells with pronounced expression of cytotoxic-activity (PRF1, GZMB), immune-cell homing (CXCL13) and exhaustion markers (HAVCR2, LAG3), and CD4 + T-cells characterized by expression of T-helper-1 (IFNG) and follicular-helper (BCL6, CXCR5) markers. In pre-treatment biopsies, the relative frequency of immunoregulatory dendritic cells (PD-L1), specific macrophage phenotypes (CCR2 or MMP9) and cancer cells exhibiting MHC class I/II expression correlated positively with T-cell expansion. Conversely, undifferentiated preeffector/memory T-cells (TCF7, GZMK) or inhibitory macrophages (CXCR3, C3) were inversely correlated. Collectively, our data identify various immunophenotypes and associated gene sets that are positively or negatively correlated with T-cell expansion following anti-PD1. We shed light on the heterogeneity in treatment response to anti-PD1 in breast cancer.