A CITE-seq analytical pipeline for the identification and molecular characterisation of dual targeting CAR T-cell products

Abstract

The clinical application of T-cells genetically engineered to express a chimeric antigen receptor (CAR) targeting CD19 is challenged by CD19 negative relapse in B-cell acute lymphoblastic leukaemia (B-ALL). Dual (CD19 and CD22) CAR targeting is thus currently under investigation, but its clinical efficacy remains unclear. We therefore compared single vs dual CAR T-cells analysing single-cell transcriptomes and CAR surface proteins from healthy donor- and B-ALL-derived in vitro CAR T-cell products for both CAR T-cell populations. Specifically, we: i) identified the different subsets, single and dual, in the CAR T-cell products without the need of prior sorting; and ii) characterised them in terms of T-cell biology, effector functions, activation, proliferation, exhaustion and cellular composition (e.g., memory phenotype). To do this, we identified CAR subpopulations using CAR transgene or protein expression, or both. We found that the combination of gene and protein CAR expression is superior, limiting the impact of RNA and protein dropout from shallow sequencing and CAR internalisation upon antigen stimulation, respectively. Moreover, by performing differential gene expression and gene-set enrichment analyses, (CAR) T-cell-related gene signature scores and cell type prediction, we detected relevant differences between the CAR T-cell subsets identified. We thus provide a computational pipeline able to identify and describe in vitro CAR T-cell subsets within the same manufacture product. This will be used for the investigation of serial samples pre- and post-infusion in the clinical setting, opening the door to dual targeting of B-ALL to fight patient relapse.