Zara Baig & Isabella Withnell: Tumor-infiltrating CD27-IgD- regulatory B cells suppress cytotoxic CD8+T cell responses in renal cell carcinoma

While more is known about the phenotype and function of circulating B cell subsets, comparatively less is known about tissue-resident B cells and how their phenotype and function are shaped by the tumor microenvironment (TME). Tumor-infiltrating B cells (TIBs) within tertiary lymphoid structures (TLSs) express high levels of MHC class I and II molecules, facilitating antigen presentation to T cells with cytotoxic activity1–4. Additionally, antibody-producing plasma cells (PCs) can mediate tumor cell destruction through antibody-dependent cellular cytotoxicity (ADCC) and complement activation5–8. B cells produce both pro- and anti-inflammatory cytokines, further regulating immune-tumor responses9. The latter function has been postulated to be exerted by regulatory B cells (Bregs), an immunosuppressive subset that may promote tumor progression by secreting pro-tumorigenic cytokines, including IL-10 and TGFβ, inhibiting cytotoxic T cells and inducing regulatory T cells (Tregs)10–14. The maturity and cellular constitution of TLSs have been linked to prognosis outcome, with mature TLSs generally associating with favorable prognosis across many solid cancers, including lung, ovarian, melanoma, and head and neck cancers15–18.

Pan-cancer atlases have begun to elucidate the diverse functional states of TIBs, distinguishing between putative pro- and anti-tumorigenic subsets that differentially influence patient outcomes across cancer types19–21. While these studies offer detailed analyses of B cell subset composition, they are often dominated by cancers with high B cell infiltration, potentially undermining functionally relevant distinctions in cancers with fewer B cells. One example is renal cell carcinoma (RCC), which exhibits lower B cell infiltration compared to other cancer types such as colon and lung cancer20. RCC represents a group of kidney cancers originating from the epithelial cells of the nephron in the renal cortex22. RCC accounts for approximately 2% of all adult cancers, with clear cell RCC (ccRCC) being the most common subtype, representing 75-80% of cases23,24. RCC mortality remains a global health concern, with over 140,000 deaths reported annually24. While early detection greatly improves survival, it is often incidental, and advanced-stage RCC is associated with significantly worse survival23. Here we provide a comprehensive characterization of B cells in RCC, identifying a population of DN1 Breg enriched in the tumor and tumor margin compared to matching blood and background kidney (BK). In vitro, we show that DN B cells suppress IFNγ expressing CD8+T cells differentiation. Together, our work offers implications for better understanding immune escape mechanisms and designing B cell subset-specific, targeted immunotherapies.

Results

Independent clustering of B cells across cancer types reveals an enrichment of DN B cells in RCC

To investigate how different tissue microenvironments and varying levels of B cell infiltration drive the differentiation of resident B cells, we compared four cancer types, breast carcinoma (BRCA), colon adenocarcinoma (COAD), lung cancer (LC), and RCC, representative of mucosal and non-mucosal tissues. To preserve cancer-driven clusters, we applied single-cell Variational Inference (scVI) to separately integrate and cluster the available single-cell B cell datasets for each cancer type (Table S1A, Figure S1A for schematic describing the pipeline). In large-scale benchmarks, scVI consistently ranks among the top integration methods, correcting both technical and biological variability, and generates a denoised expression matrix that enhances detection of low abundance genes25. After filtering out low-quality cells and excluding plasma cells to preserve better B cell subset granularity, a total of 25,427 single-cell B cell transcriptomes were obtained from 13 BRCA patients, 28,054 from 15 COAD patients, 26,652 from 15 LC patients, and 18,782 from 9 RCC patients (Table S1A).

Clustering of the scVI-derived embeddings identifies 14 transcriptionally distinct TIB clusters in BRCA, 13 in COAD, 14 in LC, and 15 in RCC (Figure 1A). We trained an additional scVI model to project B cells from each cancer into a shared latent space to identify cancer-specific and shared clusters (Figure S1A). To quantify transcriptomic similarity, we calculated cosine distances between the centroids of each cluster. Low cosine distances between nearby clusters in the shared latent space reflect similarities in underlying gene expression (Figure 1B). Hierarchical clustering based on the cosine distances organized B cells into different groups (Figure 1B). We then evaluated known B cell marker genes to validate these groups and define their biological identity.

Read full article here