Single-Cell RNA Sequencing of CAR-T Cell Infusion Products Reveals ‘viral Mimicry’ As Potential Resistance Mechanism

The CD19-targeted chimeric antigen receptor T-cell (CART19) therapy has achieved significant breakthroughs in treating relapsed/refractory large B-cell lymphoma (rrLBCL). However, the variability in durable responses among patients underscores the need to understand the molecular mechanisms of therapeutic response and resistance.

We used single-cell RNA sequencing (scRNA-seq) data of infusion products (IP) from 85 rrLBCL patients treated with CART19, encompassing 432,770 cells, to explore the heterogeneity of CD4+ and CD8+ T cells and their association with clinical response. 228,136 CD4+ T cells were categorised into 17 clusters, validated by classical markers (CD3D, CD3E, CD4, CD8A, CD8B). These clusters were distributed among patients and clinical response groups, showing individual variability in proportions, indicating that these clusters represent the majority of CD4+ T cell heterogeneity in the CAR T-cell IPs. Using established markers and gene set enrichment analysis (GSEA) of previously published T cell subpopulation profiles, we classified these clusters into five groups: naive, helper memory, cytotoxic, regulatory, and other T cells in a relatively active proliferative state. The relative proportions of CD4+ T cell types did not significantly differ between responders (complete response (CR), n=27) and non-responders (NR, n=27). We identified 204,634 CD8+ T cells in five major subpopulations across 14 clusters: naïve, central memory (Tcm), effector memory (Tem), unclassified memory, and other T cell subclusters in a relatively active proliferative state. The proportions of CD8+ T cell types were not significantly different between CR and NR groups, except for a higher proportion of STMN1+CDK1+ CD8+ T cells in the CR group. No significant differences in the proportional distribution of most CD4+ or CD8+ T cell subclusters between the clinical response groups have been observed, corroborating similar findings in recent reports.

We therefore explored whether the functional patterns of CD4+ and CD8+ cells were associated with resistance to CART19. Functional enrichment analysis using genes and gene sets from the Reactome, MSigDB, and KEGG databases revealed consistent patterns: in the NR group, various metabolic pathways, spliceosome, cell-cycle, and RNA-processing pathways were downregulated in both CD4+ and CD8+ T cells. Conversely, immune signalling pathways (e.g., interferon alpha and beta signalling), viral infection, antiviral response, extrinsic apoptosis, and dsRNA sensing pathways were upregulated. A broader metabolic pathway enrichment analysis showed upregulation of glycolysis, tricarboxylic acid (TCA) cycle, oxidative phosphorylation, and nucleotide metabolism in the CR group, while lipoic acid and linoleic acid metabolism were upregulated in the NR group. Notably, the upregulation of immune signaling and antiviral signaling pathways in poor responders to CAR T therapy is not coincidental. Additionally, in line with previous studies, RNA-processing and cell-cycle pathways were downregulated in response to spliceosome inhibition. These findings align with the “viral mimicry” phenomenon, a cellular state of active antiviral response triggered by endogenous nucleic acids typically derived from aberrantly transcribed endogenous retrotransposons or accumulated intron-retained RNAs. Based on the data and findings, we hypothesize that spliceosome inhibition in T cells of NR group increases endogenous dsRNA, triggering antiviral signaling, immune surveillance, and extrinsic apoptosis. This likely accelerates CAR T cell death, contributing to the poor persistence and therapeutic resistance of CART19. Subsequent analysis revealed upregulated expression of genes encoding antiviral signaling proteins, dsRNA-binding proteins and dsRNA sensors, and initiators of extrinsic apoptosis in CD4 and CD8 T cells of the NR group, supporting our hypothesis.

Our study highlights the heterogeneity of CD4+ and CD8+ T cell subsets in CART19 therapy and suggests a molecular mechanism of therapeutic resistance based on the “viral mimicry” phenomenon. These insights are crucial for improving the responsiveness and durability of CAR T therapy. We anticipate future research to further validate and expand these results.

No relevant conflicts of interest to declare.