Single-Cell Profiling Reveals Contribution of IFITM1 Expression to Severe CRS in CAR-T-Treated B-ALL

Background: CD19 CAR-T cell therapy has shown great promise in relapsed/refractory B-lineage acute lymphoblastic leukemia (B-ALL) patients, even though there is unpredictable variability in the occurrence and development of cytokine release syndrome (CRS). The mechanisms behind these divergent outcomes are not well understood and the heterogeneity of B-ALL patients in the aspect of both infusion products (IPs) and bone marrow mononuclear cells (BMMCs) may contribute to this knowledge gap. To explore this question, we performed a single cell transcriptomic analysis of IP and BMMC samples in B-ALL patients treated with CD19 CAR-T.

Method: A total of 120,780 IPs and baseline BMMCs were analyzed from 4 patients by unbiased mRNA profiling via single-cell RNA-seq (scRNA-seq) using the 10x Genomics system. Downstream analysis was performed using the Seurat R packages. A 13-patient validation cohort was established for result verification using bulk RNA sequencing and polymerase chain reaction (PCR) detection. To further verify at the cellular level, we knocked down the marker gene obtained in the sequencing analysis in both Nalm6 (Nalm6-KO) and THP1 (THP1-KO) cell lines respectively. CAR-T, Nalm6, and THP1 cells were co-cultured for 2 and 4 days to observe the effect of marker gene expression on the release of IL-6 in the interaction system.

Result： To eliminate the interference of disease burden (DB), the 4 scRNA-seq patients were divided into high and low DB groups in pairs. And in each group, there was one severs CRS patient and one control. The GO/KEGG analysis demonstrated that the transcriptomic profiling of both BMMCs and IPs were differentially expressed in predicting severe CRS and the enrichment of cytokine-mediated signaling was more obvious with BMMCs. To link specific cytokines to the development of severe CRS, significantly altered cytokine-related signaling pathways (n=44) were given marks listed in figure 1 - mostly IL-1, IL-2, IL-4, IL-6, IL-7, IL-8, IL-12, IL-15, interferon (IFN), and tumor necrosis factor (TNF). IFN-related pathways, including IFN-α/β/γ production, response, and mediated signaling pathways, dominated the contest (n=11). IL-1-associated pathways scored 7, while other cytokine-linked pathways had marks between 1 and 4. Similar enrichment pathways were obtained in the validation cohort.

To further clarify the cellular components that played a major role, BMMCs and IPs were sub-grouped according to cell types. Notably, differentially expressed genes (DEGs) were distributed differently between the high and low DB groups: B-ALL and hematopoietic stem and progenitor cells (HSPCs) were significant contributors in the high DB group and were perceptibly enriched in IFN and IL-1-related pathways. In the low DB group, NK/T and myeloid cells were the key donors and were augmented primarily in the IFN, IL-1, IL-8, and IL-12 pathways. All DEGs in the cytokine-related pathways from B-ALL and HSPC cells in the high DB group and from NK/T and myeloid cells in the low DB group were selected for protein-protein interaction network construction. Notably, IFITM1, a member of the interferon-inducible transmembrane family, played a vital role in every gene set. And in the validation cohort, the IFITM1 gene expression was significantly upregulated in the bulk RNA sequencing of severe CRS patients (log2FC=2.28, adj P=0.04). For cell experiments, the concentration of IL-6 released by the IFITM-knockdown THP1 and Nalm6 was significantly decreased compared with the control (FC=0.601 and 0.681 respectively). Thus, our data suggest that IFITM1 gene expression could be one of the promoting mechanisms of severe CRS during CAR-T therapy.

Conclusion： Single-cell transcriptomic sequencing highlights divergence in the BMMCs and IPs, which in concert may influence the development of CRS. Our integrated data analysis indicates the expression of IFITM1 before CAR-T cell infusion could influence the release of IL-6 and then impact the degree of CRS. Thus, IFITM1 could be a target for early intervention in high-risk CRS assessment patients.

No relevant conflicts of interest to declare.