A Single-Cell Transcriptomic Atlas of Cytokine Storm in CAR-T Therapy, COVID-19, and SLE

Introduction: Cytokine storm (CS) is a severe hyperinflammatory disorder characterized by the overproduction of diverse inflammatory cytokines, manifesting as high fever, vascular leakage, nausea, hypotension, and even multi-organ failure. CS can occur as a complication in various conditions, primarily chimeric antigen receptor (CAR)-T therapy, coronavirus 2019 (COVID-19), and systemic lupus erythematosus (SLE). These represent the three major triggers of CS: iatrogenic causes, infections, and autoimmune conditions. Given the heterogeneity of CS triggers, the clinical features and immune landscapes of CS vary widely, necessitating tailored management strategies. However, the detailed immune profiles of CS across these diverse settings are not well characterized, hindering the identification of common and distinct mechanisms underlying this hyperinflammatory complication.

Methods and Results: Here, we constructed a large-scale single-cell transcriptomic atlas of peripheral blood mononuclear cells (PBMC) in CS with data from three representative disease settings: CAR-T cell therapy, COVID-19, and SLE. The total sample count is 416, encompassing 2,304,606 cells. We collected 111 samples from multiple myeloma (MM) patients who received anti-BCMA CAR-T treatment and experienced CS events at our center, with ethical approval and informed consent from all participants. The scRNA-seq data for COVID-19 (n=221), SLE (n=40), and healthy donor controls (n=44) were retrieved from online open-access publications. Samples from the progression stage of each disease showed significantly elevated expression of CS-related genes, indicating an inflammatory status. We further analyzed the enriched cellular subclusters and expression profiles of these samples to reveal the unique CS landscape of each disease. The results showed that CS following CAR-T therapy was strongly associated with actively proliferating CD8+ T cells, with gene expressions enriched in T cell function-related pathways, such as MAPK and NLR signaling pathways. We also identified an increased percentage of CD8+ exhausted T cells in CS following CAR-T therapy, which might significantly contribute to the abnormal inflammatory responses. Moreover, COVID-19 progression was characterized by unique dynamics of myeloid immune cells, particularly CD14+ monocytes with S100A8 expression and neutrophils with enhanced interferon signaling. Interestingly, in myeloid immune cells from the COVID-19 progression stage, KEGG analysis revealed significantly enriched expressions of genes involved in neuronal transduction, suggesting the potential crosstalk between systemic inflammation and neuronal functions. In SLE, translation activities and ribosome functions were significantly elevated in B cells and plasma cells. Additionally, myeloid immune cells in SLE were enriched for genes involved in phagocytosis, antigen processing, and antigen presentation, showing the over-active status of phagocytotic monocytes/macrophages, which contributes to systemic inflammation.

Significance: Our large-scale single-cell transcriptomic analysis of CS in three distinct disease contexts uncovered the unique and shared immune landscapes underlying CS across different conditions. Our findings highlight the distinct contributions of specific immune cell subclusters and signaling pathways in each disease context, suggesting potential targets for tailored therapeutic strategies and advancing our understanding of this complex hyperinflammatory complication.